An AI generated image with at least seven errors

If we measure intelligence the way we have for the last 2500 years, humans are on the verge of losing their place as the smartest thing on the planet. We have to adapt to that and we best do it soon since the machines are growing in size, capability, and influence faster than any technology in history.

We have looked at how generative AI works in previous posts, but for context it is worth repeating a couple of things. Since the advent of the internet, we have been so inundated in information that we cannot help but miss gems of knowledge hiding in the mountain of data. Our brains have more connections and are more efficient than current foundational models, but it is probably only a matter of time before the machines catch up. Terry Sejnowski estimates that AI will have operating systems comparable to the human brain by 2050, and he is one of the more conservative prognosticators.

In the most general terms, AI takes vast amounts of information and breaks it into elements and assigns values to characteristics such as size, color, location, atomic weight, shape, age, frequency, and a few billion others. It looks for relationships between those elements and weights the value of those relationships depending on the task it is working on. Because it is working from a larger data set than any human possibly can and because it is so good at categorizing and finding relationships in data, AI is the ultimate polymath. We judge the value of modern societies based on their ability to produce polymaths.^[1]

There are, however, fundamental differences between AI and the human brain. AI acts entirely on the basis of probabilities and can be reasonably accused of producing knowledge without understanding. It has no values of its own, and, with the exception of a few rules programmed into it, lacks taste, morality, and ethics. In addition, although the machines can mimic experience learned from their training, they have no experience of their own. Daniel Dennett says the human ‘phenom’ is divided into experiences of the external world from the five senses; experiences of the internal world such as fantasy, day dreaming, mental images, recollections and ideas; and experiences of emotion (pain, hunger, thirst, anger, joy, hatred, lust, pride, awe, and regret). The machine has none of those. Let’s look a bit more closely at how those work in a lateralized human brain.

In previous posts, I have touched on the differences between left and right brain function, but I want to look at that in more detail starting with how we learned what we know.^[2] The first clear proof of lateralized function was speech which is almost always in the left hemisphere. That was followed shortly by demonstration that the left brain controls the right side of the body and is the primary recipient of sensation from that side as well. Those discoveries along with a few other lateralized functions were the result of seeing what happened when specific parts of the brain stopped working because of local damage from tumors, strokes, or trauma.

Beginning in the 1950s, Canadian neurosurgeon Wilder Penfield made a set of brilliant observations while operating on the brains of more than 500 patients with epilepsy. He used electrodes to stimulate the brains of patients who were awake and able to talk to him while they were being operated on and found that he could get reproducible reactions from stimulating specific brain areas. The motor and sensory reactions were no surprise, but he also found he could generate vivid images and memories, and most of those came from stimulating parts of the non-dominant—usually right—temporal lobes.

A decade or so later, neurosurgeons who wanted to be certain which hemisphere was dominant before operating on it injected short acting anesthetics into the carotid arteries of awake patients, literally putting half the brain to sleep while leaving the other half awake. One could find out what an unimpeded non-dominant hemisphere was doing. Even more recently, imaging techniques such as functional MRI and magneto-encephalography highlight areas of the brain that are metabolically active during specific activities.

Finally, there are some behaviors that occur preferentially on one body half or the other so one can infer the side giving rise to those behaviors.

Putting everything together, we know a good deal about how the left brain works. Speech, reading, writing, and control of the right side of the body are pretty straightforward. The left brain also is a categorizer. It breaks the world into bits and arranges them by how they relate to one another to predict what comes next. It focuses in a literal sense in that it preferentially handles information from the high resolution retinal fovea that mediates central vision. In a general sense, it looks at the world in its details and the order in which those details occur. It does arithmetic and holds specific bits of semantic knowledge and builds stories from that data. It also invents bits to fill gaps in those stories to keep them coherent. If all that sounds very much like AI, it should.

The right brain is quite different. Instead of building a picture of the world from Legos of information, the right side takes in a whole picture and reacts to it. Instead of building a picture of the world bit by bit, the right side’s world view pops into existence fully formed.

The right side responds to visual stimuli from peripheral vision the left side mostly ignores, but that is only the beginning. The right side maps the physical world for you, and it allows you to perceive and manipulate shapes. It is also the musical side. Perception of timbre, pitch, and melody all live in the right brain.^[3] There is even a part of the right brain involved in speech when it is associated with music. It is not uncommon for people totally unable to speak because of a left brain lesion to be able to sing the words of a song. Tone of voice, rhythm, stress and emphasis all come from the right side. The left side has words, but the right side gives them color.

The right side is also a hallucinator most active during REM sleep, day dreaming, or the Greek visions of the gods described by Julian Jaynes. The goddess Athena conversing with Odysseus in Christopher Nolan’s The Odyssey is a delightful example of that.

The right side is expert at humor and metaphor, both of which are difficult for the left side. The left side cannot tell a joke from a lie and fails to recognize sarcasm. It is very good at integrating related concepts, but the right side can make metaphors out of unrelated bits of knowledge. It can be argued that understanding science, philosophy, and poetry is almost completely dependent on their metaphors.

The right side is also where emotion and social interaction live. Prosopagnosia (lovely word) is the inability to recognize faces and occurs in people with large right hemisphere lesions. Those with smaller lesions in the right superior temporal gyrus have difficulty reading facial expressions and have a hard time telling what others are thinking. There is recent evidence that the communication deficits in schizophrenia and autism are actually unbalanced right brain function. Even in intact people, the left face is more expressive than the right, so we subconsciously tilt our heads during conversation to concentrate on the left side of the person we are talking to. Women—even those who have never had children—preferentially cradle infants with their left arm so the baby is in the visual field served by the right hemisphere.

Much of the memory of our experiences and especially our emotional response to those memories is in the right brain. That hemisphere takes in sensory information, filters it through past experience, and creates a picture of what is coming next. That is a level of complexity unavailable to AI, and the deep nuclei of the right side are particularly active during an ‘Aha!’ moment of inspiration.

The two hemispheres do communicate with one another, and we should look at that. The majority of interhemispheric connections pass through the corpus callosum, a bundle of 300-800 million nerve fibers connecting neurons of one side with those of the other. Interestingly, most of those connections are inhibitory; that is to say one side suppresses the other. The right side has more wiring between its neurons than the left and is easier to inhibit. In order to function maximally, the left side has to quiet down the right, and it is quite good at that.

Since Plato, Western society has placed the greatest value on the dominant side and directed education accordingly. Modernism and the age of a science-based world are triumphs of the left hemisphere while, at least since the Romantic era, the skills of the other side have been pushed into the background. Especially in the last half century information has been valued over knowledge, and skill and judgment have been secondary to quantifiable, repeatable processes. Scale, speed, and precision are favored over quality. All of that is to say that we are in a left hemisphere world, but that world is built around what AI does best. Going forward, we may need a different approach.

When I trained in neurosurgery, board certification required a two-part examination. The written part assessed semantic knowledge—the facts that live in the left brain—and we could take that part of the exam any time. I passed it six months into my residency. I then spent the next six and a half years getting experience, training my right brain, and learning to be a neurosurgeon.

In 1959, Marshall McLuhan suggested we were entering an age in which children would spend less time studying printed books and more time in apprenticeships. He was a bit ahead of his time, but I think he was prescient. To work with AI, we need to optimize what makes human learning unique, and much of that lives in experiences and the emotional reaction to them that reside in our brain’s right side.

Our Bayesian brains do not react to what is going on in real time. If they did, the processing time between sensing the world and reacting to it would leave us forever a step behind. Instead, we take sensory inputs and use our right brains to combine them with memory and emotion and make predictions. Then we act on our estimate of the likelihood that those predictions are accurate.

I have said before that being smart in a left brain world involves putting bits of information together. Genius, on the other hand, is coming up with fully formed insight—think Heisenberg’s uncertainty theory, Einstein’s theory of relativity, or quantum mechanics. ^[4] That is the right brain’s forte. To get the most out of generative AI, perhaps we should rely on it to do what it does best and concentrate on training our right brains to do what AI cannot.^[5] To that point, much of what we lump under the liberal arts involves right brain skills.

Chat GPT can take bad writing, clean up the grammar, and make it average. It can also take good writing, check it for accuracy and completeness, and make style suggestions that turn B+ work into an A. Quality is what happens when taste and experience are added to basic information. Art history majors are suddenly in great demand because they have the taste and training to write AI prompts that generate complicated virtual reality images. Anthropic has a C suite level philosopher charged with building ethics into Claude, and those only scratch the surface (if you will excuse a metaphor).

The ways AI can be leveraged have barely been touched, but one thing seems clear. AI is great at regurgitating and categorizing semantic knowledge and making associations between bits of information. It is not as good at analogy and metaphor, two of the great strengths of the right brain. Knowing art, history, and philosophy can bring those facts to life. A hybrid composed of the machine’s left brain together with the human right hemisphere is a whole new world of intellectual power. We are rapidly training the machine brain to be better, and the better we train the parts of our brain the machine cannot match, the more powerful our hybrid species will become.

^[1] This is taken from an article by the notoriously polymathic Henry Kissinger.

^[2] Although this post refers specifically to left and right brains, it is important to understand that only some of the functions have specific areas in the brain identified as serving them. Nonetheless, there is good evidence that even the functions without a specific ‘center’ are lateralized.

^[3] Professional musicians are an interesting exception. It seems that, when one is trained in music theory, the left brain begins to play a large part in sensing and making music.

^[4] Note that all of this has dealt exclusively with Western intellectual activity. Eastern philosophers have been less likely to be categorizers and intuition has a greater value in Eastern thinking. Oddly, in Eastern philosophy the two forces, yin and yang, take sides. Left sided yang is generally felt to be superior to the right sided yin.

^[5] For a more complete (and much longer) look at the right and left brains, I highly recommend Iain McGilchrist’s The Master and His Emissary: The Divided Brain and the Making of the Western World. New Haven: Yale University Press, 2009.

Illustration courtesy of ChatGPT

We are in the midst of a spasm of angst (a lot of sequential consonants in those words) concerning whether there is a long term future for the human brain in its competition with A.I. Are there things the brain can do that the machine cannot, or is artificial general intelligence the end of a 2 million year run of human intellectual supremacy on this planet?

Warren McCulloch

The age of neural networks started in 1943 when Warren McCulloch and Walter Pitts published their paper “A Logical Calculus of the Ideas Immanent in Nervous Activity.” McCulloch and Pitts were characters almost as complicated and certainly more interesting than that title. McCulloch studied philosophy at Haverford, a tiny liberal arts college outside Philadelphia, before finishing his degree at Yale. From there he went to medical school at Columbia and started a life-long study of the brain’s physiology, finishing his career at MIT’s Research Laboratory of Electronics. Pitts was as different from McCulloch as it was possible to be. He taught himself Greek, Latin, and calculus as a young child. He ran away from a dysfunctional home at age fifteen and settled in Chicago where he studied philosophy but never earned a degree. Pitts had read Bertram Russell and Alfred North Whitehead’s insanely complicated Philosophia Mathematica when he was only twelve and, when he pointed out several of the book’s logical errors to Russell, he was invited to study at Cambridge.

Pitts met McCulloch in 1942, and the two developed a mathematical theory of neural networks that led to their seminal paper. It was published when Pitts was only eighteen. Pitts eventually moved to MIT where Norbert Weiner became his mentor. When the two had a falling out in the mid 1950’s Pitts became a reclusive alcoholic. He continued to write but stopped publishing and, before he died at age 46, he destroyed his entire collection of articles. McCulloch died the same year.

Johnny von Neumann encountered McCulloch and probably Pitts as well at the Macy Conferences on Cybernetics in the late 1940s. He was fascinated with their description of neurons that integrated multiple inputs into single outputs and acted with computer-like logic. McCulloch and Pitts’ neural networks could potentially perform ‘and’, ‘or’, and ‘nor’ functions just like Claude Shannon’s relays. Von Neumann recognized that parallel neural networks had the potential to be vastly more powerful than the sequential serial programming of his computer model, but he also recognized that his hardware was neither fast enough nor well enough synchronized to mimic the brain. Besides, their memory was far too small.

McCulloch and Pitts did not deal with it, but the brain has another important level of complexity. Von Neumann’s computers were binary. Everything was coded into either a 1 or a 0. The brain is analog. A neuron can receive thousands of inputs with variable—even adjustable—strengths. Some make the neuron more likely to fire and some make it less likely to do so. The brain cell integrates all the inputs and either fires or it does not. The likelihood of firing is weighted depending on the sum of the inputs. Computers are serial and binary. Brains are parallel and analog and vastly more complicated.

Von Neumann gave up on neural networks and went back to his digital machines, although he came back to them years later when technology started to catch up. Unfortunately by that time he was dying, and his manuscript for The Computer and the Brain was only published posthumously. Von Neumann did not live long enough to see what technology could accomplish. Processors that could take billions of inputs simultaneously and adjust their weights in a way analogous to the way neurons integrate variable effects resulted in machines that could amass and correlate amounts of data many orders of magnitude greater than a human brain can and do it 10,000 times as fast. Because the machines can adjust those weights as they are trained, A.I. can learn. People who really understand A.I. predict that artificial general intelligence—a machine superior to the human brain—is only a matter of time, and maybe not very much time, but before we declare our brains obsolete, it is worth a few comparisons.

In the first place, data centers striving to achieve AGI take up acres, consume billions of watts of energy, and are cooled with millions of gallons of water. Your brain weighs three pounds, moves around on your shoulders, and runs on 20 watts. Modern neural networks have millions of processors and billions of connections. The brain has almost 100 billion neurons and a million billion connections. Imaging every cell and every connection in the human brain would require two zettabytes of data, more than there currently is in the entire world wide web. And brains are awesomely efficient. A large language model has to be trained on 1,000 times as many words as a human infant before it can converse at a similar level.

But it is not just a difference in brute compute. It is hard for a human brain to compete with A.I.’s ability to deploy semantic knowledge. The machine is a whiz with data, but does that define intelligence? One could argue that a better definition of intelligence is the ability to explain what is observed, predict what will come next, and ultimately control outcomes. Much of the way we do that is based on intuition and experience.

A very simple example is the fact that major league batters contact a pitched ball between 75 and 80 percent of the time.[1] A 95 mph fastball takes about 400 msec to get from the pitcher’s hand to the plate. To get the visual image of the pitcher from the retina to the brain and back to the muscles making the swing takes between 200 and 250 msec and the swing itself takes another 150 msec. If you do the arithmetic you can see that the decision to swing has to be made before the ball is half way to the plate. The batter is able to do that for two reasons. First, he has practiced tens of thousands of times. Second—and just as importantly—he has years of situational experience. He knows what kind of pitch is most likely given the pitcher, the catcher, and the situation.

Machines learn by using vast amounts of data to make predictions. Humans use data as well, but they augment that with a mental picture of what has happened in the past. That experience comes from a combination of memory and input from the five senses. At least for now, machines are anchored in place and their only version of experience is second hand from their training data. In addition, for A.I. every problem is new and every solution starts from zero.

Experience is not the brain’s only advantage. We have talked about the fact that connections between neurons are adjustable. A.I. does that solely by varying the weights of inputs based on the statistical likelihood of an outcome. Humans do that as well, but they have another way to change weights—emotion. The strength of a neural connection varies with how much of specific neurotransmitters is released at each synapse, but neurotransmitters also change with the mental state of the brain’s owner. Dopamine increases in pleasurable situations. Norepinephrine increases with threats. Nursing babies cause a gush of oxytocin in their mothers., and those are only a few of the chemicals the nervous system uses. Variations in mental state—particularly states involving interaction between humans—change how the brain processes information.[1] There are mental states (pain is a good example) that profoundly alter the way brains deal with the world even though they are hard or even impossible to put into words. It is hard to imagine A.I. competing in that world. Foundational models can imitate emotions but they cannot have them.

At least since Gutenberg’s press allowed most of humanity to read, we have measured intelligence as a combination of memory, literacy, and numeracy. A.I. is better at all of those than we are. However, the machines have no accumulated experience, and their ability to evaluate and predict is purely statistical. In a constantly changing world occupied by constantly changing humans, the machines are at a disadvantage.

Perhaps going forward, we should change our definition of intelligence and change the skills we train for. When we learned to write, our left brains took command and visions generated by our right brains receded. When we learned to mass produce written material, we gave up the ability to memorize and devoted our brains to creating new knowledge. For the past century or so, we have increasingly spent our time and mental energy interacting with an explosion of data and information. Now A.I. can do that for us. Perhaps we need to train for intuition and empathy.

[1] However only 5-8 percent of the hit pitches result in a safe base hit.

[1] It is also true that many of those same chemicals directly impact how the body functions. They make pupils constrict, raise the heart rate, and redirect blood flow to the muscles for example. Emotion changes the body as well as the brain.

Generative AI is the fastest growing technology in human history. More than 800 million people use it at least weekly, and data centers account for about half the growth in the United States gross domestic product.

There is no question that large language models can sort through more information than a human brain and do that orders of magnitude faster than we can. There is, however, an ongoing argument about whether we are on the verge of artificial super intelligence—a machine that can do everything a brain can do and do it better—or whether that is even possible.

To deal with that question, we need to start by defining intelligence. Think about what a large language model is. One feeds it a huge amount of information composed of words, parts of words, or combinations of words broken into tokens. The input is all language. The machine then does its statistical magic to decide which words go together to reply to a prompt which is also in language and produces an answer in—you guessed it—language.[i]

If we measure AI against the brain solely on how many facts one or the other contains and how fast each can get to those facts (semantic memory), the machines have already won. Of course, they make mistakes and make things up, but those are solvable problems. But more to the point, semantic memory is not the only kind of memory.

What about stored experience? We have a rich library of memories of the people, places, and situations we have encountered during our lives, and the library is almost all in images rather than words. Those memories help us make predictions and know what to do when we encounter similar situations.

Large language models only have the words that people have used to describe experiences. They have no experiences of their own. A number of AI experts who think superintelligence is inevitable—most prominently Dario Amodei of Anthropic—recognize that problem and offer a solution. Their “world view” comprises AI-equipped robots that can wander around seeing, hearing, and touching. The robots have all the senses of a human and remember what they have encountered. Does that mean they have experiential memory equivalent to that of humans?

Perhaps language does not fully represent experience. A situation described in the same words may have entirely different impact from one person to another. The same view from the edge of a cliff may be exhilarating to one person and terrifying to another. Metaphors are one way we account for experience (“Sky high” or “Scared to death”), and metaphor is one of the things large language models are especially bad at.

Another way to look at this is to ask whether there is something in human intelligence that takes place without words. It seems to me that the answer is yes. Consider for instance, the experience of pain. Not only do we experience pain without putting words to it, we actually have a hard time finding useful words. In reacting to pain, we resort to sounds that pre-date language—“Ouch” or “Arghh”—or expletives that are still sounds without meaning. As a surgeon one of my most difficult tasks was getting a reliable sense of a patient’s pain. We had a short list of descriptors (stabbing, throbbing, burning) and an inadequate set of measures (one to ten or various frowny and smiley faces) but the risk of over or underestimating the amount of pain or not figuring out where it came from was a constant source of worry and a frequent source of error.

Much of what makes experience powerful is the emotion associated with it, and that emotion is entirely personal to the person having the experience. In addition, it is intricately tied to an endocrine system that humans have and machines lack. We release oxytocin in the presence of human attraction and norepinephrine when we are scared for instance, and the endocrine response is involuntary and entirely personal.

Beyond that, we can all generate images in our brains that are independent of language. There have been recent experiments using electrodes implanted on the brain’s surface in people who have lost the ability to generate speech. Those brains generate specific, reproducible patterns when the people are asked to imagine doing a specific activity. Just the activity—not the words that describe it. Think about that. You have no trouble generating an internal picture of a thing, a place, or a situation that is quite clear and uses no words. A large language model can describe that internal world, but it cannot reproduce it.

Brains can create metaphors, they can generate emotion, and they can paint ineffable pictures. Generative AI can do none of those things.

And that brings us to the last point. If we are to make the best use of generative artificial intelligence, we need to maximize what we can do that the machine cannot. Much of that may be in the realm of ideas outside the bounds of language. That is the sort of right brain creativity we have undervalued at least since we learned to write, more since the Enlightenment, and even more since we digitized information.^[ii] Perhaps we will have a different measure of intelligence as we learn to work with AI.

[i] I am only dealing with large language models here, although the same argument would apply to AI that generates images, music, and so forth.

[ii] My thanks to Sean Trott and his Substack “The Counterfactual” for helping put this into words.

“Speech is a river of breath, bent into hisses and hums by the soft flesh of the mouth and throat.”

                                                                                                                        Steven Pinker

Poetic, but not quite that simple.  That river comes from the lungs through folds in the larynx (“vocal cords”) and then to the back of the throat.  If you are one of the click speakers like the African Khoisan, you can generate a popping sound with the soft palate.  If you are the rest of us, you change the air flow by pushing your tongue against your hard palate, “k” or hard “c” and in candy in the back of the tongue, “g” as in good a bit more forward, “d” as in dog in the middle, and “t” as in toy at the front.  You can use similar tongue positions to end a syllable—hike, hog, and hid.  You can also leave a bit of space between the tongue and palate and push air through with a hiss as in show or “s” as in similar if the tongue is a bit farther forward. You can also move the air all the way forward to your lips to get ‘m’ as in many or “b” as in board.  Vowels are air movements that change in tone as air is pushed through.  Try it. 

English speakers can typically make sounds by opening and closing the mouth three to eight times a second to produce an average of five vowels and five consonants, but the sounds are not all there is.  The meaning of the sounds changes with pitch—Really? Really!!—accompanying gestures, or the context from words around them. 

The number of vocal sounds (phonemes) varies from language to language.  English has 44, Khoisan has 141, and the Rotokas of Pacific Bougainville have only 11.  Did those oral gymnastics arise just so we could talk?  Probably not.  Exaptation is the idea that we use anatomy (and brain circuits as well) that evolved to do one thing (eat or breathe) to do something else (speak), but why were humans the only animals to make that adaptation?

To answer that, we can start with our primate ancestors, the closest of which are chimpanzees (Pan troglodytes) and the closely related bonobos (Pan paniscus). Genus Homo split off from them around six million years ago and comparing us to them makes the questionable assumption that genus Pan has not evolved in that time.  Nonetheless, 96 percent of our DNA sequences are the same, and it is worth looking at how close to speech non-human primates have come.  An intriguing bit is that chimpanzees, bonobos, gorillas, and orangutans all show a right hand preference, although significantly less strongly than humans.  That said, Broca’s area, which you will remember is crucial to generating speech, is six times as large in humans as it is in chimps.  Chimpanzees do have a repertoire of calls signifying warning, territorial claims, and the like, but the calls are entirely holistic.  None breaks down into parts and most originate from the emotional tracts of the limbic system and the brain stem, not from the frontal lobes.  Chimps do not have the same repertoire of sounds as humans, and the sounds they do have lack the complexity of real speech.

The oldest known transition from apes to humans was Sahelanthropus about six million years ago.  The next step, Australopithecines (one of which is the A. afarensis nicknamed “Lucy” who has been called the mother of humanity) were bipedal and lasted from about 4 million to 2 million years ago.  Australopithecines had only about one-third the brain volume of later Homo species, and it seems unlikely that they had much more speech than earlier primates. 

Next in line were Homo habilis who lived from 2-2.5 million years ago. Whether they had speech—or if they did what kind--is unknown, but H. habilis were tool makers, and flakes left from their toolmaking suggest that they were made with a right hand.  Their Broca’s area is significantly larger than that of chimpanzees even though their whole brain only measured from 550-800 cubic centimeters.

The next step brings us closer to real speech. Homo erectus emerged in Africa about 1.9 million years ago and was the first of our human ancestors to move out of that continent.  H. erectus was in Asian Georgia around 1.8 million years ago, in Java and China 1.6 million years ago, and may have been in southern Europe 1.5 million years ago.  They overlapped on one side with H. habilis and on the other almost to Homo sapiens, lasting until just 110,000 years ago.  Besides being travelers, H. erectus were formidable tool makers.  Their innovations (Acheulean stone tools) included bi-facial hand axes and cleavers.  The argument that a species that lasted close to 2 million years, that built villages, that designed tools, and that traveled over half the planet must have had speech seems plausible. Although we cannot prove they spoke, it is hard to imagine passing those skills from generation to generation without language.

Before we go any farther, it is necessary to deal with an alternative theory for the origin of speech.  Massachusetts Institute of Technology and University of Arizona professor and linguist Noam Chomsky has vehemently argued that human speech came as a result of a spontaneous genetic mutation about 60,000 years ago that caused the brain to suddenly grow a language center. That theory led to a decades-long search for a language gene that has proven almost entirely futile. The current academic consensus is that language gradually evolved over tens of thousands of generations and that there is no single gene responsible for its evolution.

The last evolutionary step was Homo heidelbergensis that branched into several species (H. neanderthalensis, H. floresiensis, and the Denisovans) that overlapped with H. erectus and Homo sapiens and even interbred with the latter.   After emerging between 550,000 and 750,000 years ago, all except Homo sapiens became extinct about 40,000 years ago, and it is likely they all possessed some degree of speech.  Neanderthal brains averaged about 1,400 cubic centimeters, some 10 percent larger than those of modern humans, and the genetic makeup of Neanderthals and modern humans is strikingly similar.  There are only 96 identified genetic mutations responsible for proteins that differentiate H. sapiens from their predecessors.  One of those produces transketolase-like protein 1 (TKTL1) that indirectly increases the production of neurons in the frontal lobes.  It is not the ‘language gene’, but TKTL1 probably has a role in making the brain capable of speech.  The important thing to remember is that the brain went from about 500 cc to 1,300 over three million years.  There was ample time in 125,000 generations for evolution to occur.

The oldest Homo sapiens fossils date to about 300,000 years ago and have been found throughout the African continent.  Their finely made flint blades and points were attached to handles and shafts for the first time.   By 100-200,000 years ago, H. sapiens had spread across Asia and Europe. Their brains were modern size—four times the volume of a chimpanzee brain—and had especially large frontal lobes.   H. sapiens could talk; now let’s look at how that happens.

We have looked at how speech might have arisen based on culture and brain size.  Another way to look at the problem relies on the “ontogeny recapitulates phylogeny” trope. The basic idea is that what an organism goes through as it matures mirrors what happened when its species evolved.  The problems with that are obvious, but it is still worth considering what happens with human infants and children as they learn to talk. 

First, no one is born knowing language; that led to a couple of intriguingly harsh experiments.  The Greek historian Herodotus claimed that Egyptian king Psamtik I (664-610 BCE) locked two newborns in a shepherd’s hut with no exposure to anyone but the hut’s mute owner.  The idea was that the first words the infants spoke, having been exposed to no words, would reflect the original language of the human race.  One of the infants was thought to have uttered “bekos”, the Phrygian word for bread, convincing Psamtik and succeeding generations that Phrygian was the first language.  King James I of Scotland tried a similar experiment with Scottish infants whom he said emerged speaking “very guid Ebrew.”

Besides Phrygian and “Ebrew”, the number of possible languages for an infant to learn is not small.  There are currently more than 7,000 languages spoken on Earth, although 96 percent of the population speak only 4 percent of those.  Fewer than 1,000 people speak 1,750 of them, and one (Resigora spoken by Arawaks in Amazonian Peru) has a single speaker.  Indo-European, which gave rise to 400 languages including all European languages, Bengali, Kurdish, and Farsi, appeared around 4500 BCE.  Every one of those had to be learned.

The rate at which children learn to speak is staggering.  The average twenty-four-month-old starts out learning two to four words a day, and by the end of the second year is learning ten to twenty. The average five-year-old has a 10,000-word vocabulary. 

There is a sequence in which children learn that may well reflect how speech originally developed.  Children learn iconic words that sound like what they name (smash, boom) followed by those that name objects or shapes.  Next come simple actions those objects do— “Mommy go.” Paul Elkman has suggested that we are born with a basic set of emotions—anger, sadness, happiness for instance—and we learn to attach words to those feelings early on.  Later, children learn to understand and voice what the people around them want.  This understanding of beliefs and desires (the theory of mind) may be the specific deficit of those on the autism spectrum.  Children then learn that concepts including physical entities, relationships, emotions, and ideas come in hierarchies—Some dogs are poodles; some poodles are miniatures; my miniature poodle is Frazier. 

Chomsky argued that anything as complex as language had to be entirely inherited—it was just too complicated to learn—but he was thinking of the brain as a serial processor in which each step had to be completed, and the result determined the next step.  Terrence Sejnowski recognized that was wrong.  The brain operates with multiple simultaneous processes in parallel and adjusts its output based on prior experience and statistical likelihood. The more it learns, the better it gets.  That training has one counterintuitive anatomic correlate. At birth, each neuron has about 2,500 synapses.  By three years of age that is trimmed to about 1,500, and in adults it is cut in half.  As the brain is trained, it becomes more efficient.

So where are we with all of that?  First, what we know about how speech started is by implication only. We have no direct evidence of events for which we have no written record.  Second, speech probably came slowly and over a long enough time to be part of evolution. Third, even though some of the mechanisms for speech come from our DNA, we still have to learn language, and that learning fundamentally changes our brains.

There is one more thing. Of the seven classical liberal arts (Grammar, rhetoric, logic, arithmetic, music, geometry, and astronomy) rhetoric is the cornerstone, and for good reason.  Speech was the first and remains the most common way for one person to inject the contents of his or her brain into the brain of another.  The better one is at rhetoric, the more likely it is that injection will change the listener’s brain. There is a compelling reason to train one’s brain to communicate effectively.  Of course, now speech can come from non-human brains and we are subject to having our brains changed by artificial intelligence.

References:

Chomsky, Noam, Language and Mind. New York: Harcourt Brace Jovanovich, Publishers, 1972.  A dominant theory of the origin of language for much of the second half of the 20^th century now largely discarded.

Dehaene, Stanislas, Reading in the Brain: The New Science of How We Read. New Yok: Penguin Random House LLC, 2009.  An excellent discussion of the neuroscience behind speech.

Mithen, Steven, The Language Puzzle:  Piecing Together the Six-Million-Year Story of How Words Evolved. New York: Basic Books, 2024.  The archaeology of speech.

Pinker, Steven, The Language Instinct: How the Mind Creates Language. New York: Harper Perennial Classics, 1994. A classic.

Illustrations are all from The Smithsonian Museum of Natural History Human Origins. https://humanorigins.si.edu/evidence/human-fossils/species.

Human brains steadily enlarged beginning 7 million years ago when species Homo split from the other primates (although the size has actually gone down a bit since peaking with Neanderthals).  The interesting part is that all animals evolved two brains that are almost mirror images of one another.  For the most part, the two halves of the brain enlarged in tandem although markings on the inside of ancient skulls suggests that there was some right left asymmetry as far back as Australopithecus 3.5 million years ago.  That asymmetry goes through the evolutionary scale all the way from fruit flies to modern humans in whom dominant temporal lobes are somewhat larger than the non-dominant ones.  That said, the differences are small.  Since handedness and speech overwhelmingly reside in the left hemisphere, why has the right side continued to take up so much space and consume so much energy even though there has never been a human society in which right brain dominance was common? Surely a parsimonious evolution would not have kept so much metabolically expensive real estate around if it did nothing important.

At almost the same time Paul Broca identified the left hemispheric speech area, British neurologist John Hughlings Jackson suggested a unique function for the right side.  “There are two modes of expression, one emotional and the other intellectual. By one we show what we feel, and by the other we tell what we think.”[1] Jackson saw emotional expression in patients who had lost all ability to speak. Although completely aphasic, his patients still had emotional outbursts and were especially adept at swearing.

As it turns out the right brain is essential not just for emotional reactions but for social interaction in general, and that is not unique to humans.  We need to do a quick bit of clinical neurology to go farther into that. Look straight ahead and extend your right arm with the index finger sticking out.  Now move the hand back even with your right ear.  You won’t see the finger.  Now gradually move your hand in an arc until you see the finger.  It has entered your right visual field.  Input from your eyes is arranged so that messages from the part of the left eye next to the nose merges with that from the outer side of the right eye, and they go together to the left occipital cortex to give you a coordinated view of the right side of your world.  Same on the left, and the same all the way down the evolutionary ladder.  Now you understand why animals encountering others of the same species have a strong preference for keeping them on the left.  They want them in the visual field that goes to the “social” hemisphere.

That is not the only indication that the right brain is involved in social interaction.  Nursing animals prefer to have infants on their left and human mothers have a strong tendency to hold infants in their left arm so the baby’s face is in their left visual field.  Interestingly, that “cradling bias” is present in women who have never had children and in girls as young as six years old, but it is much less common in autistic mothers who are socially disconnected.

Muscle movements involved with emotion—smiling or sneering for example—are more pronounced on the left side than the right, and people with right hemisphere lesions have less ability to read the facial emotions of others.  Studies of patients with brain injuries repeatedly confirm that human production and recognition of social stimuli are a right brain function.

The right hemisphere may have a role in species preservation as well.  Response to infants is not the only thing lateralized to the right.  Reaction to other emotional input—erotica or threat for example—is more rapid when the stimulus comes from the left visual field.  That probably accounts for the fact that animal mothers in novel or threatening situations keep their infants on the right so their left visual field is unobstructed. 

For the two brains, there seems to be a ladder of skills.  On the left, the ability to repeat patterns of motor activity evolves into the ability to manipulate and use tools which culminates in the ability to communicate.  On the right, predator avoidance evolves into recognition of emotions and culminates in social skills.  The left side lets you learn how to eat; the right keeps you from being eaten.

Social interaction is not the right brain’s only skill.  One way to control intractable epilepsy that originates in the left brain is to remove the entire hemisphere.  If that is done before the patient is six years old, the child still learns to speak.  In that limited sense, the right brain functions as a backup system.  The right brain is also involved in spatial and perceptual relations.  Another epilepsy operation involves cutting all the connections between the two hemispheres so out of control electrical activity cannot spread through the whole brain.  The result (besides seizure control) is creation of two independent brains.  In those patients, a right hander can still write with the dominant hand but cannot draw with it.  People with right brain lesions also have a great deal of difficulty mentally rotating complicated figures.  Musically, the left brain does a pretty good job with rhythm, but the right brain seems to be necessary for recognition of tone and pitch.

At least since Gutenberg, we have placed an inordinate value on what our left brains do.  We concentrate on training the speech hemisphere and we judge intelligence on how well it performs.  If we are entering a time when much of that work will be outsourced to machines, perhaps we will need to take a harder look at the other side. What would we look like if we spent as much time and effort training the right brain as we have the left?

 References:

Coren, Stanley, The Left-Hander Syndrome: The Causes & Consequences of Left-Handedness. New York: The Free Press, 1992.

Forrester, Gillian S., William D. Hopkins, Kristelle Hudry, and Annukka Lindell, eds., Cerebral Lateralization and Cognition: Evolutionary and Developmental Investigations of Behavioral Biases. Cambridge, MA: Academic Press, 2018.

[1] John Hughlings Jackson in Medical Times and Gazette” 1866.  cited in  Taylor, James, ed., Selected Writings of John Hughlings Jackson, Volume Two: Evolution and Dissolution of the Nervous System, Speech, Various Papers, Addresses, and Lectures. London: Hodder and Stoughton Limited, 1932.

We have looked at alphabets, manuscripts, printing, and coding, but we have not looked at the earliest and possibly the most important way we move information—speech.  Our ability to move air from our lungs through our vocal cords, palate, tongue, and lips and turn it into communication is uniquely human.  To be sure, other animals can make sounds that convey some information, but it is almost all at an emotional level (fear, anger, satisfaction) and deals only with the present.  We can turn complex thoughts into words that deal with the present but also with what has happened and what might happen, and speech allows us to move those ideas from one brain to another, and for almost all of human history that was the only way we could to do that.

Finding exactly how speech originated is impossible.  In the 1950s Noam Chomsky hypothesized that language started 100,000 years or so ago when one of our lucky ancestors spontaneously developed a language gene. Chomsky made a career on that theory, but the majority of those who study linguistics now lean toward language having gradually evolved over something more like a million years and dating as far back as Homo erectus. That said, there is still an active search for a genetic basis for speech. 

Proof for any theory of how speech started is hard to come by since there is no direct record going back that far.  Unlike with writing, we cannot study language development in primitive societies that do not talk because there aren’t any.  As far as we know, there has never been a culture without speech.  Indeed, there are something like 7,000 languages currently in existence and many more than that have gone extinct.

We can learn something from the way children learn language since no one is born talking.  Starting at age two, children learn an average of two to four new words a day and have an average 10,000-word vocabulary by age five.  They go from simple names for things to words for what things do to words for characteristics of objects or activities and then to complex grammar that allows a handful of sounds (forty in English) to express an infinite number of thoughts.

Where in the brain that learning takes place is not completely clear, although we do know some things. In 1861 French physician Paul Broca encountered Louis Victor Leborgne at Hôpital Bicêtre in Paris. Leborgne’s nickname “Tan” came from the fact that it was the only syllable he could utter.  When Tan died, Broca obtained the brain and found a specific lesion at the base of the left frontal lobe in an area which now carries his name. [1] 

In 1874, German neurologist Carl Wernicke described a different speech deficit with a different anatomy.  His patient could speak fluently but could not understand a word of what was said to him.  Broca’s patient could not speak but could understand; Wernicke’s could speak but could not understand.  A variety of aphasia variants have subsequently been described, but all are associated with lesions around the Sylvian fissure that divides the temporal lobe from the frontal and parietal lobes, and virtually all are on the left side.                                           

The right-left thing is interesting.  Roughly 90 percent of humans of whatever era and culture are right-handed, and in almost all right-handed people speech comes from the left peri-Sylvian region.  It does so in about 70 percent of left handers as well. [2] Language does not even have to be verbal to be lateralized; deaf people with left sided brain damage exhibit deficits in signing identical to verbal aphasia. 

One would think that anything as important as handedness or speech would have been explained, but that is not the case. At least 60 species of animals exhibit handedness. Chimpanzees exhibit clear right-handedness in their tool use.  Human fetuses as young as 15 weeks of gestation show a preference for right thumb sucking, and infants as young as two months old preferentially respond to speech from their right side.  Left brain communication is not unique to humans. Zebra finches have to be taught song patterns, usually by their fathers.  New chicks babble like human infants until they learn their songs, and their ability to sing is lateralized to a specific area in their left hemispheres.  Handedness and left dominance for communication seem to be genetically programmed across species, and no one knows why.  It might make more evolutionary sense to spread something that important around so single injuries could not take it away, but that is not how it works.

One more thing—when we speak or listen, we do not do it one word at a time.  We become fluent because we use our previous experience with language to fill in gaps where sounds are unclear and to predict what will come next before we say it or hear it.  Does that sound familiar?  It should because that is exactly how AI’s large language models create answers.  They take a word or part of a word and find the most likely next sequence based on text on which the model has been trained.  In that sense generative AI acts just like a brain.  Well like a left brain anyway.

References:

Chomsky, Noam, Language and Mind. New York: Harcourt Brace Jovanovich, Publishers, 1972.

Everett, Daniel L., How Language Began: The Story of Humanity’s Greatest Invention. New York: Liveright Publishing Corporation, 2017.

Geschwind, Norman, Selected Papers on Language and the Brain. Boston: D. Reidel Publishing Company, 1974.

Pinker, Steven, The Language Instinct: How the Mind Creates Language. New York: Harper Perennial Classics, 1994.

[1] In 1836 French physician Marc Dax may have related speech abnormalities to left brain damage, but his observation, if it was made, was largely ignored.

[2] Although if the left hemisphere is damaged or surgically removed before the patient is six years old, speech can develop from the right side.

If you remember, we started this series with Julian Jaynes’s idea that writing led to suppression of the non-dominant, hallucinating hemisphere and the loss of the gods. Later, we looked at the memories of medieval scholars.  The scarcity of books, the inconvenience of scrolls, and scripta continua made memorizing an entire text the only practical way to access what had been written, and scholars regularly carried the contents of hundreds of books in their heads.  Printing rendered memorization unnecessary, and the skill was lost in a generation. Reading, writing, and memorizing are learned skills—the brain has to be trained to do them, and that training changes the brain. To know whether generative artificial intelligence might cause similar changes, we need to look at how brains work.

            Learned skills come in three general categories.  Procedural memory—what we usually call muscle memory like riding a bicycle, typing, or playing scales—is automatic and not readily reduced to  computer code.  Episodic memory involves experiences from one’s past.  It is difficult to code directly, but it can be captured and filed as audio and video recordings.  For 13^th century scholars and for those of us in the age of AI, semantic memory is the interesting kind.  It involves things like meanings, definitions, and concepts that are not specific to a single time or place.  Semantic memory is uniquely adaptable to being stored in a database and to being coded.  It deals with facts unlike episodic memory which brings back situations and procedural memory which lets us perform learned actions without thinking.

            If you recall, memory is stored in two ways.  In the short term, the strength of synapses changes so neurons connect to one another either more or less readily.  In the long term altogether new synapses and even new neurons form.[1] In addition, different kinds of memories are stored in different places. When an external stimulus (somesthetic, auditory, visual) comes into the brain it is relayed to the cortical area responsible for processing that sort of stimulus (parietal lobe, temporal lobe, occipital lobe). Implicit memories—automatic skills and habits—are forwarded for storage in the cerebellum, the deep brain striatum, and the amygdala in the medial hippocampus to be recalled as needed.  Once relayed, the memories can held briefly or converted and saved.  In each case, training changes brain structure.

            Now let’s go back to history. Before writing, the elements of culture were preserved by bards and poets who learned society’s stories.  Rhythm, rhyme, and formula were learned and passed from generation to generation as poetry and performance saved cultural information.  Writing was a new, more efficient way to preserve that information and it freed Greek brains to write plays and to speculate about how the body, the body politic, and the world worked. Homer gave way to Aristotle.

Much the same happened in the Middle Ages.  Before printing, creativity had come to be viewed with more than a little skepticism.  Anything beyond what was passed down from the Greeks and Romans or the Church fathers and the Bible was frivolous.  Intelligence was measured by how much ancient writing one had stored in his or her (mostly his) memory.  The more you could recite, the smarter you were. Moveable type changed all that.  In a generation or two, memory was externalized—it was outsourced to printed books.  How much information you knew quickly became less important than how much you knew about where to find it.  As fast as memory declined in importance the skills associated with memorization atrophied, and brain space was freed up to discover calculus, anatomy, and astronomy.

            That brings us to the present day, but first a quick recap: information processing as speech caused anatomic changes in the brain that were passed down genetically, but that process took hundreds of thousands of years.  On a much shorter time course—that of a single human life—training changes brain circuitry.  Learning changes not only how neurons connect to one another but also how many connections there are. Parts of the brain with a host of new connections become more important and occupy more of our conscious attention.  Skills change, and sometimes those that are less important drift into the background or are lost altogether.  That brings us to AI.

            Computers are staggeringly adept at the same skills needed for semantic memory.  IBM’s Watson can process 500 gigabytes of information—roughly a million books—a second.  We can never hope to match the information and its retrieval or the ability to find relationships between bits of information that Chat GPT will have. It is plausibly predicted that general AI will have an operating system equivalent to that of the human brain by 2050.  Much of what we have prioritized our left brains to do will likely be outsourced to computers.

            So what do we do about that?  We could try to eliminate AI, but even if we wanted to that time has passed.  Besides, censorship has never worked well; witness printing, the Church, and the Inquisition.  We could try to limit the use of AI, but that just puts us at the mercy of those who choose not to impose limits. The best we can do is to keep the machines’ answers honest and try to understand what they are doing.

            The third alternative, and the one we most likely will have to take, is to partner with AI, but that will involve rearranging our brains.  What might that look like? Ulrich Kraft said, “The left hemisphere is responsible for convergent thinking and the right hemisphere for divergent thinking.  The left side examines details and processes them logically and analytically but lacks a sense of overriding, abstract connections.  The right side is more imaginative and intuitive and tends to work holistically, integrating pieces of an information puzzle into a whole.”

            How about  a metaphor or two?

            The left brain sees trees.  The right brain sees the forest.

The left brain is an accountant.  The right brain is an artist.

            The left brain is a navigator.  The right brain is an explorer.

            The left brain plays chess.  The right brain dances.

            The left brain is a journalist.  The right brain is a novelist.

            Full disclosure—I am terrible at metaphors.  Those were all from Chat GPT.

            As we have seen, written language has made our left brains dominant and suppressed the other side.  If we outsource what the left brain does, we might free conscious space for other uses much like writing let Aristotle imagine biology and printing  let Newton dream up gravity.

            There are clues that might actually happen.  People with left brain damage from fronto-temporal dementia will occasionally develop musical and artistic ability they never had before becoming ill.  Allen Snyder used transcranial magnetic stimulation to temporarily impair left brain function and the experimental subjects suddenly had new artistic skills.  It is even possible that not needing to learn the order of the presidents or capitals of states will contribute to the visual-spatial skills of video game players.

            Writing made the left brain dominant, but writing is a learned skill.  Maybe a different kind of learning will bring back the right side.  Printing made memory less important, and time spent with that skill went elsewhere to mankind’s great benefit.  Speech was associated with major changes in brain anatomy.  Although those changes took a very long time, unlike learned skills they were heritable. Maybe if we last long enough, AI will turn us into a different sort of thinking organism.

            Of course, that is all pure speculation.

Further reading:

Febvre, Lucien and Henri-Jean Martin, The Coming of the Book: The Impact of Printing, 1450-1800. London: Verso, 1976.

Gleick, James, The Information:  A History, A Theory, A Flood. New York: Vintage Books, 2012.

Havelock, Eric A., The Literate Revolution in Greece and Its Cultural Consequences. Princeton: Princeton University Press, 1982.

Kissinger, Henry, Eric Schmidt, Daniel Huttenlock, The Age of AI and Our Human Future. New York: Little, Brown and Company, 2021.

[1] For a refresher, see “Can That Machine Really Think?”

“Barring some catastrophic planetary nightfall, the tech industry will create within decades, machines with human-level intelligence and behaviors capable of speech, reasoning, highly coordinated actions in economics, politics and, inevitably, warcraft.  The birth of true artificial intelligence will profoundly affect humankind’s future, including whether it has any.”

Christof Koch

            So far we have been looking at times when changes in how we communicate led to changes in how the brain handles information.  We have looked at the development of speech, writing, and printing.  It might be worthwhile to look in more detail at how the brain has changed and to consider what might be in store for it going forward.  This is a bit complicated, so I will divide it into two posts to keep it manageable. Here is part one.

            Speech is a special case because 50-60,000 generations have allowed time for genetics to take over.  Hominin brains are vastly different from those of chimpanzees, our nearest primate relatives; our brains are four times as large and our thinking lobes have orders of magnitude more neurons. Neanderthal brains reached their maximum size about 600,000 years ago and had a genetically determined increase in the number of their frontal lobe neurons.[1] Mutations made the brain capable of speech and the ability to imagine the future, both of which are unique to humans. It is an odd side note that brain size peaked in those Neanderthals at 1,459 ± 182 cubic centimeters while modern human brains average only 1,328±145 cubic centimeters. The decrease in brain volume came only about 3-5,000 years ago—after speech but roughly coincident with the development of writing.  Our brains have either become more efficient or they lost something about the time we acquired the new technology.

            What happens in the brain when people learn to speak?  Although no one is born knowing how to talk, small children—even those with significant intellectual impairment—acquire the skill with astonishing rapidity.  Between ages two and five, children learn an average of two to four new words a day. The average five-year-old has a 10,000-word vocabulary that will grow to about 88,700 words during subsequent school years. Early childhood is the only time we can become fluent in a language (or occasionally more than one) in such a short period.  Noam Chomsky argues that is because humans are born with grammar somehow coded into our brains by a genetic change that occurred 50 or 60,000 years ago.  Most scientists think it is more likely that anatomic changes associated with speech  (for example a tongue designed to handle food adapting to manipulate sounds) slowly developed over one or two million years.

            Regardless, the brain changed dramatically as half of it became inextricably linked to speech.  The angular gyrus sits at the junction of the parietal, occipital, and temporal lobes where it is uniquely situated to bring together tactile, auditory, and visual inputs.  The major sensory inputs are processed and merge in the angular gyrus which has been justifiably called the integrator of integrators. In 1874 twenty-six-year-old Carl Wernicke proved that damage to the left angular gyrus rendered the victim unable to produce comprehensible speech, and Joseph-Jules Déjerine subsequently found that the damage to the same area led to the loss of the ability to read and write.  Language lives in the dominant angular gyrus.

            We cannot leave without briefly considering what was happening in the non-dominant half of the brain.  In the years before PET scans and functional MRI’s, I remember doing Wada tests to be absolutely sure which hemisphere was responsible for speech before surgery on the brain, and that led to some interesting insights.[2]  In the Wada test, a catheter is put in the carotid artery and a short acting anesthetic injected.  If the artery feeds the dominant hemisphere, the patient temporarily loses the ability to talk.  The interesting thing is that a fair number of those patients retain the ability to sing with the lyrics intact—words associated with music evidently coming from the non-dominant side.

            In the early 1960s Canadian neurosurgeon Wilder Penfield experimented on more than 500 patients undergoing operations most of which were designed to remove areas of damaged brain responsible for intractable seizures.  Because the brain itself has no sensation, Penfield was able to operate with the patients awake; he could electrically stimulate various parts of the brain during the procedures and observe the results.  Stimulation of the parietal, occipital, and temporal areas on the non-dominant side could produce vivid somatosensory, visual, or auditory hallucinations.  Remember Julian Jaynes and the loss of the gods?  Hallucinations seem to be a non-dominant talent, an observation that you might recall was recently verified with functional MRIs in actively hallucinating patients.

            For our purposes the important thing is that, over hundreds of thousands of years, the human brain structurally changed to allow it to communicate with words and symbols, and those changes were coded into our DNA and pass from generation to generation.

            Writing is different.  In the first place, unlike speech, there is direct historical evidence of how literacy changes the brain.  Second, the changes wrought by literacy occurred over a few thousand years, not tens of thousands of generations.  There has not been enough time to code the changes into our genes; writing cannot be handed down from one generation to the next.  The brain changes associated with writing all must be learned—they are the changes of training, not heredity.

            Nonetheless, writing does change the brain.  Non-literate communication is almost entirely auditory whereas reading uses the visual brain.  Literacy moves information processing from the auditory temporal lobe to the frontal lobes and from the non-dominant to the dominant hemisphere. Areas that were genetically designed to recognize and name objects are re-programmed to recognize letters and words.  Groups of trained neurons recognize written language and associate it with memory so quickly that it is automatic.  We read without thinking about it, but reading starts as a learned skill.

            The learning has not always been pleasant.  The oldest examples of teaching reading we have are Sumerian clay tablets with a teacher’s example on one side and the student’s attempt to reproduce it on the other.  The school days were long and learning the thousands of cuneiform characters was a burdensome chore.  The schools had a “man of the whip,” and the last words on a student tablet were often, “And then he caned me.”

            The cuneiform example brings up an interesting point.  In assigning literacy to the dominant side of the brain, we have been tacitly assuming we were talking about characters that represent sounds.  Remember in cuneiform and some Eastern writing most characters represent not a sound but a thing.  You might guess that would affect lateralization, and you would be right.  Readers of Chinese ideograms recruit brain areas across both hemispheres.  Japanese is even more interesting.  It has two writing systems—kana which is phonetic and handled with the dominant hemisphere, and kanji which uses ideograms and draws on both sides. A patient with a unilateral brain lesion can lose one type of writing but retain the other.

            This is a good place to stop for now, but there are two main things to remember.  First, the brain does physically change with acquisition of communication skills.  Second, the change can be either inborn or learned, although the genetic changes take a very long time.  In the next post we will look at the changes when information exploded and at what is happening now and might happen in the future.

References

Geschwind, Norman, Selected Papers on Language and the Brain. Boston: D. Reidel Publishing Company, 1974.

Wolf, Maryanne, Proust and the Squid: The Story and Science of the Reading Brain. New York: HarperCollins Publishers, 2007.

Wolf, Maryanne, Reader Come Home: The Reading Brain in the Digital World. New York: HarperCollins Publishers, 2020.

[1] Of about 19,000 hominin genes, only 96 have been identified that separate Neanderthals from modern humans.  One of these is responsible for producing transketolase-like protein 1 (TKTL1) that, in the laboratory setting, does increase neuron production.

[2] Almost all—but not every—right-handed person is left dominant, and half of left handers are as well.

“Knowledge is of two kinds.  We know a subject ourselves, or we know where we can find information upon it.”  Samuel Johnson

“There is a growing mountain of research. But there is increased evidence that we are being bogged down today as specialization extends. The investigator is staggered by the findings and conclusions of thousands of other workers—conclusions which he cannot find time to grasp, much less to remember, as they appear.” Vannevar Bush

            Libraries, concordances, indexes, encyclopedias.  In the last post we looked at the growth of information and the futile struggle first to collect it and then to store it, organize it, and find it. 

To call the rate at which information is being created mind boggling is an epic understatement.  I spent $10,000 to chain four external drives for my Apple IIe computer in the late 70’s so I could have a whole megabyte of data storage.  A terabyte is a million megabytes, and just a few years ago that much storage would have required a whole rack of disk drives.  Now I have it on a memory stick on my key chain.  A petabyte is a thousand terabytes.  The numbers are getting big enough to make your eyes glaze over, but, just for reference, a petabyte of storage would hold 799 million copies of Moby Dick. A thousand petabytes are added to the internet every day, and the internet now houses over a zettabyte—a million petabytes.  The few hundred thousand volumes collected by the Ptolemies in the library at Alexandria are a grain of sand on the beach.

            By the beginning of the 20^th century, hard copy libraries were already inadequate.  Computers looked like a solution in the second half of the century, but getting information into a data base required coding, and that required more than plain text and decimal arithmetic. 

When he died in 1621, English mathematician and astronomer Thomas Hariot left a clear description of arithmetic using base 2 instead of base 10.  Hariot had been sent to survey the Virginia colony and may well have written his manuscript  to pass the time during a long Atlantic crossing.  Sadly, his paper remained undiscovered until J.W. Shirley stumbled on it in 1951.

            At about the same time Hariot was writing, Frances Bacon invented an alphabet using strings of just two letters—A and B.  Bacon’s system—in which A is AAAAA and T is BAABA for instance—looks remarkably like ASCII, but there was no real use for it and the idea died of irrelevance.  In 1703, Gottfried Wilhelm Leibniz proposed using a binary system for a system of weights and coinage to be used by German governments. Leibniz credited the Chinese king “Fohy” with having devised s similar system almost 4,000 years earlier.  In the Chinese system a short dash stood for yin and longer one for yang, but they might just as well have represented 1 and 0.  As was the case with Bacon’s binary system, Leibniz generated no enthusiasm.

            Only mathematicians, engineers, and science nerds had any interest in non-decimal number systems until 1946.  As you might remember, even the Manhattan Project Eniac computer pictured above was a base 10 machine.  It took John von Neumann to realize that computer switches were either on or off and that computer logic was intrinsically binary.  It only made sense to program and store data in base 2.

The idea of using computers to manipulate data as well as store it came at about the same time.  In 1945, Vannevar Bush published “As We May Think” in Atlantic Monthly in which he described how information might be handled in the future. He included the Memex, a machine designed to store a person’s entire life including every book, record, and written communication.  All personal photographs would be digitized and the user would wear a forehead-mounted camera attached to view finder eyeglasses to film every event.  The Memex would store all of that on a giant disk attached to a desk with a keyboard and multiple buttons, levers, and screens to retrieve and manipulate the data.  None of the requisite technology existed in 1945, but by 2009 Gordon Bell of Microsoft was doing just what Bush had suggested with his MyLifeBits project.  Virtually unlimited storage along with bulk scanners, wearable cameras and microphones, and location trackers made Bush’s science fiction fact. 

            Meanwhile, Berkeley Professor Guy Montgomery had started a concordance of the poetry of  John Dryden. When Montgomery died in 1951, he left a quarter million index cards crammed into sixty-three boxes.  Josephine Miles, who was teaching English at the university, took the cards to the Electrical Engineering Department where there were IBM computers.  The alphabetized terms, poem titles, and line numbers were put on punch cards, and the first computerized index was born.

            In 1963, Susan Artandi submitted her Rutgers PhD dissertation, “Book Indexing by Computer” with support from the United States military.  She proposed indexing directly from unedited data based on a pre-selected dictionary of search terms.  The system was uniquely suited to the explosively growing and increasingly digitized scientific literature.  When I was still writing scientific articles, I spent hours going through the Index Medicus which is built on the system Artandi described.  The problem is that the system is restricted by the choice of search terms, and those are chosen by the editors. Even with the restricted terms list, every year’s edition of the Index has several columns of tiny print in multiple ponderous volumes.  Literature searches were painful and most often incomplete.

By mid-century, it was clear that there had to be a better way to access the mass of information that was being digitally stored, and the problem was about to become orders of magnitude more difficult.  On August 6, 1991 Tim Berners-Lee announced the availability of a project that “aims to allow links to be made to any information anywhere.”  Anyone interested in the project just had to drop him a note and he would forward the necessary code.  The internet was born, and the sorting problem became a crisis.

            In 1996, two Stanford graduate students—Sergey Brin and Larry Page—were working on Backrub, a project aimed at ranking search results more effectively than alphabetization or counting how often a search term appeared on a web page.  They came up with PageRank (eponym intended) that ranked search results based on how many times the result had been used by other sites and how important those other sites were.  That is much like the standard practice of ranking an academic paper based how many times it is referred to in other articles.  Ranking by relevance was a revelation.  Brin and Page realized the economic opportunity and left Stanford to found Google, and the idea of ranking by relevance to a specific task was picked up by GPS, social media, and Netflix among others.

            So how does computer storage and retrieval of information compare to what goes on in the brain? Humans have three kinds of memory.  Semantic memory includes meanings, definitions, and concepts and is what is typically contained in databases.  Procedural memory covers skills like typing, riding a bicycle, or playing musical scales and is mostly not programmable. Episodic memory preserves past experiences that do not typically go into databases but can be saved in audio and video recordings.  Computers are best at mimicking semantic memory, and they do that best when retrieval is done in context.  Ranked retrieval was an immense advance, but real contextual retrieval like Chat GPT is a revolution.  As Henry Kissinger said, contextualized information becomes knowledge.

            Humans are incapable of looking at all the information stored on the internet and they are orders of magnitude less able to find relations between parts of that information, but today’s fast computers can do all that.  IBM’s Watson has 16 trillion bytes of random access memory and can process the equivalent of a million books a second.  Generative AI can scan a database with trillions of elements, find relationships among those elements, weight them, and predict an outcome in a few seconds.  The machines do that on a scale incomprehensible even to their programmers.  In traditional programming, each step is clear and can be examined by the programmers.  In generative AI what goes on in a transformer’s multiple layers is invisible, and programmers regularly have no idea why one outcome is preferred over another.  The current revolutionary data analysis cares nothing for causation.  It is only interested in correlation.

            And that fundamentally changes our relation to information. The great philosophical leap of the Enlightenment was that a hypothesis that explained an observation turned it into knowledge.  Data was collected in an experiment.  A hypothesis was formed to explain the results and then used to predict new results.  If the predictions proved accurate, the hypothesis was assumed to be correct.  That knowledge could then be used to control outcomes.  Nature could be explained without resorting to magic or faith, and in many cases could be controlled.   

Generative AI is a different animal. We can collect and store more information than any human can possibly look at. The new AI has given us a way to manipulate that data in ways we cannot understand, and that results in an unbridgeable gap between knowledge and understanding.  The computer tells us what it has learned, and we often have no idea how it learned it. We just have to trust the machine.  In a very real sense, we are going from the rationality of the Enlightenment back to and age of faith;  it is just that the higher power is a computer rather than a deity.

References.  Here are a few of the books I relied on most heavily.

Duncan, Dennis, Index, A History of a Bookish Adventure from Medieval Manuscripts to the Digital Age. New York: W.W. Norton & Company, 2023

Garfield, Simon, All the Knowledge in the World: The Extraordinary History of the Encyclopedia. New York: HarperCollins Publishers, 2022.

Glaser, Anton, History of Binary and other Nondecimal Numeration. New York: Tomash Publishers, 1971.

Kissinger, Henry, Eric Schmidt, Daniel Huttenlock, The Age of AI and Our Human Future. New York: Little, Brown and Company, 2021.

O’Gieblyn, Meghan, God, Human, Animal, Machine: Technology, Metaphor, and the Search for Meaning. New York: Anchor Books, 2021.

Sejnowski, Terrence J., The Deep Learning Revolution: Artificial Intelligence Meets Human Intelligence. Cambridge, MA: The MIT Press, 2018.

By 8500 BCE Sumerians settled between the Tigris and Euphrates Rivers in what is now southern Iraq. Annual flooding brought fresh soil to replenish the delta so grain could be grown year after year without depleting the land.  Farmers produced more than they needed to sustain themselves and their families, so they could sell the excess.  Trade required a record—information that could be entered, saved, and retrieved, and long before other records were written economic data was being collected.  The Sumerians made quarter-sized lumps of clay and inscribed marks that stood for the name of a commodity and how much of it was represented—five goats, two sheep, six jugs of wine.  The tokens were collected and saved in tennis ball-sized clay “purses” that recorded ownership and transactions in the oldest extant form of data storage and retrieval.

                Five thousand years went by before the Sumerians created a set of symbols that represented words rather than quantities, and it was another fifteen hundred years before Hammurabi had his laws inscribed on a seven-and-a-half-foot piece of black stone.  At first the accumulation of information was slow, but by the seventh century it had picked up.  In 1849, at the site of the ancient Assyrian capital Nineveh, British archaeologist Austin Henry Layard and his assistant Hormuzd Rassan made an astounding discovery.  They found the two main chambers of a library largely collected by Ashurbanipal (668-627 BCE), the last major Assyrian king.  Tens of thousands of clay tablets inscribed in cuneiform were stored on shelves in cabinets arranged by topics that included not only administrative records and laws but also medicine, astrology, history, and a variety of chants, prayers, and stories.  Each cabinet was given an identifying letter, and the shelves were numbered from the bottom up so new tablets could be piled on as needed.

Cuneiform is not easy to read, and the library arrangement made it likely that only the librarian  knew where things actually were.  As far as we know, Ashurbanipal’s library marked the first time the volume of information threatened the ability to find it.  Cumbersome, poorly organized libraries were fine for preserving information but were poor places to do research.  That was not as significant a problem as one might think since scribes were essentially the only ones who could read, and they spent their time copying what had already been written rather than trying to create anything new.

                Farther west, clay tablets inscribed with styli were superseded by marks made with ink on either papyrus or vellum sheets that were glued together and rolled up.  The scrolls could hold a huge amount of information—50-foot-long rolls were common.  One scroll that documented the rule of Rameses II stretched to 133 feet. There were no chapters and no page numbers, and the title of the book and its author were almost always at the end of the roll rather than the beginning.  The scrolls were stacked on shelves or, if a single volume needed more than one roll, in buckets on the shelves.  Although the scrolls lacked titles, some did have tags with a title hanging out—σίλλλυβόι, hence syllabi.

                The problem with retrieving information from a scroll was that, besides their being bulky,  one had to unroll the whole thing to find a passage unless the volume had been committed to memory, so the text was mostly used to verify what the reader remembered.  As in Assyria, Egyptian libraries were places to preserve what was already known with little intent to add anything new.

                The greatest of the ancient libraries was founded at Alexandria during the reign of Ptolemy I Soter about 288 BCE.  Tradition has it that Demeterius of Phaleron wanted to succeed Aristotle as head of the Athenian Academy but did not get the job.  Instead, he went to Egypt when Ptolemy offered him the chance to start a great library.  Demetrius came with no small amount of ambition;  his initial goal was to collect everything that had been written which he estimated would be about 500,000 books.  From the beginning, he bought some books and copied others.  For decades any ship that docked at Alexandria was forced to turn over every book on board.  What happened next varied.  Sometimes the original was returned; sometimes just the copy went back and the original stayed in Alexandria.  Ptolemy II put down a 15-talent deposit (equivalent to one skilled workers wages for seven years) to borrow a definitive copy of the works of Aeschylus, Sophocles, and Euripides from Athens.  When it came time to return the book, he forfeited the deposit and kept the original. 

                The Alexandrian collection may not have had the full 500,000 volumes, but it was huge and encompassed vast parts of Greek literature that no longer exists.  The library was organized by subject and topics within the subjects, but scrolls were undoubtedly hard to find and certainly hard to research if they were found.

                Over centuries the library  at Alexandria suffered from neglect and from repeated assaults on the city.  Some of the collection may have been lost in the 3rd century CE When Aurelian’s forces burned part of the city.  One contested theory is that the library was finally destroyed when Amr ibn al-As took the city in the 7^th century.  It is said that he decreed that all books in the library that did not deal directly with Islam should be burned.  Islamic literature that was not the Koran was not necessary and should be burned as well.  It is said that he distributed the scrolls to local bath houses to be used as fuel. At any rate, the library vanished along with an irretrievable treasury of ancient knowledge.

Romans copied the Alexandrian Greeks.  Augustus founded two libraries in Rome—one in Greek at Campus Martius and another similar in size and design for Latin on Palatine Hill.  Tiberius, Vespasian, and Trajan all built libraries in Rome, and Hadrian built one in Athens.  For the Roman emperors, libraries were status symbols.

An important change came in the early centuries of the Christian Era.  Codices—individual pages stitched together and put between covers—had been suggested by the Roman poet Martial in the 1^st century.  He praised a friend for folding parchment pages into leaves—a codex—instead of rolling them into a scroll.  Martial pointed out that codices were smaller and more appropriate for the traveler than bulky rolls, but his suggestion never caught on.

It was left to early Christians to popularize the form.  A codex held six times the content of a scroll; all four gospels fit in a book that would have taken five rolls.  Although they were not paginated, it was easier to flip pages to find a passage than to unroll an entire scroll, and the Christians placed a high value on having accessible liturgical writing. Besides being cheaper and more portable,  codices were easier to copy.   As long as manuscripts were hand copied however,  pagination, even in codices,  was impossible.  No two copies were exactly alike.  

During the Middle Ages, universities began collecting collect books that were not just stored but used for research.  By the 13^th century Oxford, Cambridge, and the Sorbonne all had reference libraries, although important books like dictionaries, concordances, Biblical commentaries, and the works of Thomas Aquinas were chained to desks.

Printing brought a fundamental change to information storage and retrieval.  As we saw earlier, there was an explosion of books in the first century and a half after Gutenberg’s Bible.  Ten to  fifteen thousand new texts were printed in the first fifty years, and 150-200 million books came out of the presses in the 16^th century.  Throughout the Middle Ages libraries had almost all been in religious or educational institutions.  Now private individuals could own books.  By the end of the century, “haberdashers, weavers, drapers, tanners, grocers, cheese mongers, hawkers, locksmiths, pastry cooks, skinners, dyers, shoemakers, and watchmakers had libraries.”  Gottfried Leibniz, David Hume, Johann Goethe, and Immanuel Kant all spent time as librarians. Books were affordable and abundant.  The increase in available information was staggering, but retrieval had become a huge problem.  The amount of information gushing out of printing presses far outstripped the ability of researchers and librarians to find it. 

One solution was to make it easier to search individual volumes.  As we saw in an earlier post, ancient Ugarit, a city in what is now Syria, developed an alphabet of consonants between 1400 and 1300 BCE.  The revolutionary thing about the Ugaritic system is that the letters were in strict order, something not true of either hieroglyphics or cuneiform.  That “abecedary” passed on to proto-Canaanite and to the Greeks, and eventually the orderly alphabet was used to create indexes.  In fairness, it was impractical to index hand copied scrolls or even codices since pagination varied from copy to copy.[1]  Nonetheless, rough alphabetization had been used in Alexandria.  The librarian Callimachus sorted the library into areas of learning and used an alphabetical list of the illustrious men hung above each shelf as an indication of what was there.  He also compiled a catalogue of what was in the library that took up 120 scrolls.

Sorting knowledge remained a serious problem through the Middle Ages.  Robert Grosseteste, a British poet, polymath, Chancellor of Oxford, and Bishop of Lincoln made a valiant attempt to categorize everything he had ever encountered.  Around 1230, Grosseteste divided what he had learned into groups of similar ideas.  He then divided each topic into one of three areas—Biblical, Church fathers, and pagan literature.  He created a system of 400 symbols that he drew in the margins of books to indicate what was in the text.  None of his books were paginated so it was necessary to scan each page looking for the right symbol to find a passage, but Grosseteste had created a search engine, albeit a clumsy one.

At about the same time, Cardinal Hugo de Saint-Cher enlisted 500 monks to sort through the Bible which had recently been divided into books and chapters (although not verses).[2]  Each monk was assigned a letter or part of a letter and sent to find every word that started with that letter and record where that word was found.  The alphabetized list of 10,000 terms would prove invaluable to local clerics writing sermons.  The concordance was a term based search engine and a major advance in information retrieval and presaged Google.

Printing eventually changed the world of data retrieval in another way; every page of a printed edition is precisely the same from volume to volume so the pages can be numbered. The first paginated book was printed in Cologne just fourteen years after the Gutenberg Bible, and tables of contents and indexes promptly followed.  Given the avalanche of printed material, it was just in time.  Still, indexes were only good for the individual volume, and the number of editions was staggering.  At the Second Conference of Librarians held in London October 2, 1877, J. Ashton Cross, the librarian of the Oxford Union Society, suggested a Universal Index of Subjects.  The goal was to compile an index of every branch of knowledge—essentially an index of indexes with each library to provide an index of its holdings and contribute a guinea a year to fund the project.  Cross’s great index proved too ambitious and it fizzled in less than a decade.

Another alternative is to condense information into a single volume or set of volumes.  The oldest known attempt at an encyclopedia (from the Greek enkyklios paideia or learning within a circle—literally well-rounded education) was the aggressively titled Assyrian All Things Known in the Universe from the 2^nd millennium BCE.  Plato’s nephew Speusippus published an encyclopedia in 370 BCE, and Marcus Terentius Varro published a nine volume summary of grammar, rhetoric, geometry, astrology, and music  between 16 and 27 BCE.  Unfortunately none of those survived.

Pliny the Elder’s Naturalis Historia was interrupted by the eruption of Mount Vesuvius that killed him in 79 AD.  He had already published ten books of a project that ultimately came to thirty-seven volumes completed by his nephew also named Pliny.  In the aggregate, it is the largest single work from the Roman Empire to survive.  It is divided into topics including Astronomy, Geography, Physiology, Zoology, Botany, Pharmacology, and Minerology.  Together, the Plinys claimed to have consulted 2,000 books by more than 1,200 authors to accumulate in excess of 20,000 facts and observations.

Pliny’s encyclopedia was unsurpassed until 1255 when Vincent of Beauvais published Speculum Maius (The Greater Mirror) , 80 books with 988 chapters and 4.5 million words divided into four sections.  Speculum Naturale dealt with astronomy, anatomy, agriculture, and light in the order in which they were created according to Genesis.  Speculum Doctrinale covered medicine, the mechanical arts,  and theology. History from the creation to the Last Judgment was in Speculum Historale.  Speculum Morale was added after Vincent’s death and is mostly taken from the writings of Thomas Aquinas.  In all, it was the largest western book until the mid-1500’s.

Europeans were not the only ones dealing with an excess of information.  Zhu Di, the 3^rd Emperor of the Ming Dynasty who ruled between 1402 and 1424 and who built the Forbidden City, ordered all Chinese knowledge to be gathered.  The resulting Yongle Dadian comprised 22,937 sections in 11,095 volumes bound in yellow damask.  Since it was written in Chinese ideograms, organization was by topic and there was no index.

The Enlightenment made a catalogue of rapidly growing western knowledge imperative.  Denis Diderot undertook the job and, between 1745 and 1772 compiled and edited (along with Jean Le Rond d’Alembert) an Encyclopedie that eventually came to 17 volumes containing 71,8128 articles accompanied by 3,129 plates in 11 more volumes.  The Parlement of Paris suppressed the work after the first seven volumes in 1758, but with the assistance of government Director of Book Trade La Moignon de Malesherbes who warned Diderot and hid his papers, the project was surreptitiously finished along with supplementary updates and a two volume index.

Not to be outdone, the British published their own three-volume Encyclopedia Britannica between 1768 and 1771 in Edinburgh.  Like the Encyclopedie, the Britannica was sorted by topics arranged alphabetically.  The work is a cautionary example of information based on the interests and biases of the editors.  Farriery (treatment of the diseases of horses) got 40 pages.  Morals merited another 40 and the Law 76, but Murder got only 4 lines.   Surgery warranted 38 pages, Midwifery 39, and Medicine 112 more.  Brasil (sic) got 12 lines while Norway got only 7, the first 5 of which were latitude, longitude, and surrounding countries.  The rest of the entry was brief:  “It is a cold and barren country.”[3] Panama got 5 lines also comprising the location to which was added:  “Where the treasures of gold and silver and other rich merchandize of Peru are lodged in magazines till they are sent to Europe.” 

Americans got into the act as well.  Thomas Dobson began advertising subscriptions for an American version of the Britannica in 1789, but he added a number of local contributors and dropped Britannica from his title before the encyclopedia was published.  One of his  authors, who wrote entries on individual state histories, American Indians, and Revolutionary War battles, was a man we have encountered before.  Jedidiah Morse was Samuel F.B. Morse’s father.

Britannica’s high point was the 11th edition published in 1910.  My own copy is 28 volumes of double column small print on tissue thin pages that occupy an entire library shelf. The 11^th edition has been justifiably called the last time it was possible to put all western knowledge in a single set of books.

Libraries concordances, indexes, and encyclopedias were all attempts to manage an exponentially increasing amount of information.  By the last half of the 20^th century they had all proven woefully inadequate.  An entirely new approach was necessary.  Fortunately it appeared, and that will be where we will pick up the story in the next post.

[1] The Torah was an exception.  Exact copying was a religious obligation and the Torah scrolls were identical from one to another.  Regardless, they were not indexed.

[2] The illustration at the beginning is the first known image of a person wearing glasses, although it was painted more than a century after Hugo’s death.

[3] The 11^th edition of the Britannica had twenty-two double column pages that would have run to something more than fifty pages in the same type and page set as the first edition.

There seems to be little question that the new version of artificial intelligence and large language models like Chat GPT will change how we interact with our information sources, and that got me thinking about another time when the way information was transmitted changed.

            We have already looked at the evolution of the alphabet.  Pictograms were graphic representations that bore no relation to a language’s sounds. Logograms, which have persisted in Asian languages, were more abstract but still not phonetic, and the brain handles logograms differently than it does a phonetic alphabet.  Japanese still has two separate writing systems—kanji which is logographic and kana which is phonetic and syllabary.  Brain injured Japanese patients brought up with both systems have been carefully studied; kana uses the left brain exclusively while kanji needs both the right and left sides.

The earliest Semitic alphabets did represent sounds but came without vowels and initially represented the way syllables sounded.  Sumerian and Egyptian writing, besides lacking vowels, were an uninterrupted string of characters, some of which were read right to left, some left to right, some top to bottom, and some first in one direction and then the opposite one—boustrophedon from the Greek for oxen that plowed a field back and forth.  The text had to be read aloud to make sense, and it was best if the reader already knew what he was reading since the code left many gaps.  The ancient syllabary texts were better as reminders of what one already knew than as sources of new information.

            In an effort to make the code easier to decipher, the symbols were divided, usually with periods or spaces, but the division went away when the Greeks added vowels.  Greek was written in scripta continua, a constant string of letters with no separation between paragraphs, sentences, or words.  If one is reading left to right across the page (as the Greeks did) the eye’s ability to focus extends about fifteen characters out.  It does not take much imagination to see how difficult it might be to decipher what is meant by an uninterrupted river of characters, only fifteen of which you can concentrate on at a time.

            Why the Greeks chose to do that—and why Latin writers eventually adopted scripta continua as well—is unclear.  One suggestion is that continuous script simply packed more words onto an expensive parchment.  Perhaps the Greeks just liked how it looked on the page.  Maybe reading was intentionally made difficult.  As we noted earlier, the twenty-two Greek characters were much easier to learn than cuneiform or hieroglyphics, and perhaps those who made a living reading and writing made literacy harder to maintain their corner on the market.  All of those are guesses, but the end result is not.  Scripta continua profoundly affected how manuscripts were reproduced and read.

            The scriptoria where manuscripts were reproduced were a constant undercurrent of sound.  The representations we have of the copyists show them sitting with pen in hand and manuscript balanced on a knee. There was no space for a source from which to copy.  The scribes were being read to by a lector while the writer got the words on a page in block-like, perfectly regular script.  The reader was a performer, but the scribe was somewhere between a technician and an artist, and scribes were valuable.  In medieval Ireland, the penalty for killing one of them was the same as that for murdering a bishop.

            The important thing here is that reading scripta continua almost had to be done out loud; silent reading was an anomaly.  When St. Augustine went to study with Bishop Ambrose of Milan in the late 4^th century, he was amazed to find his teacher alone in his cell with a book.  “When he read his eyes scanned the page and his heart sought out meaning, but his voice was silent and his tongue was still.  Anyone could approach him freely and guests were not commonly announced, so that often, when we came to visit him, we found him reading like this in silence, for he never read aloud.”[1]  There had been scattered references to others silently reading—Euripides, Demosthenes, and Julius Caesar—but it was uncommon enough to always be remarkable.  In fact, in Hebrew and Aramaic a single word is used for both speaking and reading.

Things changed in the early 7th century beginning with monks in Ireland who probably wanted to make the Bible more comprehensible.  They reintroduced the separation of Latin into individual words.  From Ireland, individual written words spread to France and Italy in the early Renaissance, to Slavic Cyrillic in the 17^th century, and to Sanskrit only recently.

                       Once words were separated the look of the page changed drastically. A stream of characters was broken with spaces, capitalization, and decoration. Most importantly, it was no longer necessary to read text aloud to decipher it and reading fundamentally changed.  Silent reading made it possible to digest the complex ideas of Albertus Magnus and William of Ockham.  Clerics and educated laymen  could study alone and even put their thoughts in glosses and margin notes.  (The thoughts were not entirely religious.  There was a small explosion of both homosexually and heterosexually explicit marginal illustrations in religious literature.)  Text read and contemplated silently broke the Church monopoly on the written word and gave rise to heretical ideas that were anathema to Rome, and some of those ideas made it into tracts that were publicly circulated.  Christian heresy had been around as long as there had been a church, but since written thoughts had to be read out loud, it was risky and never too great a threat.  Silent reading made it dangerous, and the first heretic was burned at the stake in Orleans in 1022.

            Separating text into individual words fundamentally changed how reading was done and how information was transferred.  That small change shook the status quo. Imagine what a large language model might do.

References:

Manguel, Alberto, A History of Reading. New York: Viking, 1996.  Very readable and a wealth of arcane and interesting information.

Saenger, Paul, Space Between Words: The Origins of Silent Reading. Stanford, CA, Stanford University Press, 1997.  Scholarly impressive research, but a bit dense for casual reading.

[1] From Confessions, vi, 3.

Celsus de re medicina (1497) Both paper and print remain perfect.

On May 26 the Wall Street Journal reported that Nvidia was on the verge of joining the likes of Microsoft, Apple, and Alphabet as a $1 trillion company because of the furor set off by Chat GPT and a bevy of other artificial intelligence companies.  Nvidia does not do artificial intelligence.  The company mostly made chips that powered computer games until those chips turned out to be uniquely suited to generative AI.[1]  The company’s value has exploded because it has a key piece of technology. 

Something similar happened when Johann Gensfleisch (whom we know as Johannes Gutenberg) made a better moveable type and put it together with other technology that already existed in the middle of the fifteenth century.  His printed books were the first identically mass produced items, and language was mechanized. To understand how that happened, we need to start with a bit of history.

It is generally accepted that Gutenberg’s 1454 Bible was the first book printed with moveable type and that the Mainz goldsmith was uniquely responsible for the technology behind it.  Both are untrue.

The oldest known printed book currently resides in the British Museum.  It is a Buddhist scroll printed from wood blocks in 868 CE.  Xylography, another name for reproducing images with carved stamps, spread through most of East Asia long before Gutenberg.  The Haein Temple in South Korea has 81,258 wood blocks used to print versions of the Buddhist Tripitaka between 1236 and 1251. Asia had more than xylography; in 1377 a Korean anthology of Buddhist teachings was printed with metal moveable type. 

In spite of that, printing never became an industry in the Far East. Part of the failure was due to strict control of a process governments considered dangerous, but it was also the result of the Asian script and the attitude about writing.  In the first place tens of thousands of Chinese ideograms made moveable type a formidable undertaking, but the fact that hand drawn calligraphy was seen as an essential fine art was at least as important.  How a manuscript was drawn was at least as important as what it said, so printing was left to the Europeans.

Four bits of technology were necessary for the printing press to work—paper, ink, moveable type, and the press itself. We can start with paper.  Had the Gutenberg Bible been printed on vellum as most manuscripts had been, it would have required 15,000 skins to produce one edition of 100 volumes. That was obviously impractical.

Paper had been used in China for an astonishingly long time.  Chinese tradition holds that Han Dynasty official Cai Lun (ca. 50-120 CE) invented paper making after watching wasps build their nests.  He was said to have pounded the inner bark of the mulberry tree mixed with hemp fibers to make a paste that could be dried in sheets.  In fact, fragments of ancient paper made from hemp have been found in a tomb near what is now Xi’an, although they were blank and probably used for wrapping rather than writing.

Paper was brought to Europe by Arab traders, and by 1268 paper was being made in Umbria. A number of Italian mills followed soon after. In 14th century Genoa paper was used by notaries, although more important documents such as royal charters and government decrees continued to be written on more durable parchment.  Paper making spread with the cultivation of flax and linen. By the mid-1400s, there were paper mills in France, Germany, the Low Countries, Vienna, Cracow, Copenhagen, and Buda. By the 16^th century, mills in France alone were turning out 450,000-900,000 reams of paper a year and paper makers became rich.

Between the 14^th and 17^th centuries, the paper making process hardly changed at all.  Rags were collected by specialized dealers and delivered to water-driven mills, often converted from grain processing. The rags were pounded by water-driven mallets studded with nails or small knives.  The resulting pulp was suspended in warm soapy water to make a paste and laid on a wire frame to drain. After an initial drying, the paper was put on felt to draw out the last water, pressed several times, and hung to dry.  Next it was coated with sizing to make it less absorbent when ink was applied before a final polishing with flint.  Finally, the sheets were collected in 25-sheet reams that were bound in 50-ream bales for delivery to market. 

The whole process was water intensive.  A typical mill consumed 700 liters of water/hour/kilogram of paper.  The water had to be fast running to drive the mill and almost perfectly clear to keep from staining the paper.  Good quality cloth was needed as well, and white linen underwear was preferred although, as printing spread, rag shortage became a chronic problem.[2] There is a certain irony in the fact that the spread of knowledge was reliant on used undergarments. Mills first tried making thinner paper, and then printers made smaller books to preserve the precious fiber. Two pages printed on each side and folded in half made four pages a sheet, a folio.  The number of pages steadily increased to four a side (octavo), then twelve (duodecimo) and sixteen (sextodecimo).  Readers complained about the shrinking size and printers about the decline in paper quality.  By the 17^th century regulations that sound like current computer chip restrictions limited paper export or banned it outright. Countries that had imported paper were forced to build their own mills.

Ink was less of a problem.  The water-based ink used on vellum manuscripts did not penetrate paper well and did not stay on the fonts.  Although we have no samples of Gutenberg’s ink except what is left on the printed pages, it is clear the early printers quickly developed a thicker, oily black paste that stayed on the letters and permeated the paper sheets. Red cinnabar was used to “rubricate” initial letters of pages and paragraphs.

The fonts were more challenging. Wooden and clay fonts had been invented in Asia, and King Sejong of Korea had even ordered 100,000 sorts (individual characters) to be made of copper in 1403, but it took the metallurgic knowledge of European goldsmiths to perfect the process.

A font (or fount) is a collection of all the characters needed to reproduce a language and includes lower and upper case letters, punctuation, symbols, and numbers.  The first fonts were made to look as much like hand drawn manuscript letters as possible although there was no standard size for fonts until the 18^th century when a “point” was defined as 1/144 the size of the French king’s foot.  First, each character is carved in reverse onto a hard metal—originally brass or bronze and later steel.  This “punch” is then hammered into a softer metal—usually copper—to make a mold.  Then a third metal—typically a mixture of lead, tin, and antimony—that melted at a lower temperature than the copper was poured in the mold.[3]  The result was thousands of copies of the meticulously carved punch.  The individual characters or sorts were filed as closely as possible to the same height so they would all strike the paper when mounted in the frame for pressing.

A printer would typically own 60-100,000 sorts which allowed him to set up a few dozen pages after which the letters were retrieved for the next few pages.  A single page of a folio Bible in the 16^th century would have required 5,000 sorts (10,000 to do both the front and back of a page), and, if all the pages of the 1,000-page Bible had been kept and stored, it would have required 15 square meters of storage space and would have weighed more than 2.5 tons.

Before we get on to the press, a word about how the type was set.  The compositor sat on an uncomfortable stool at a slightly tilted desk in front of a cabinet with rows of pigeon holes, each of which contained copies of one sort—small letters in the lower case and capitals in the upper case.  At first every printer had his own arrangement of the sorts, although the layout was eventually standardized so compositors could go from shop to shop.  Eventually the bottom of each sort was notched so the compositor who knew where each letter was stored could pick up what was needed and put it in the slot of a composing stick in the correct orientation without looking, much the way a typist uses a keyboard. When the line on the composing stick was complete, it was placed in a tray—the galley—and a lead spacer was placed below it to keep sequential lines separate.  When the page was complete, the lines were secured in the galley and the whole thing was ready for the press.  In a typical twelve to sixteen hour day a good compositor was said to be able to turn out 2,500-3,000 pages. One can imagine a hunched compositor never looking at his flying hands as they filled the sticks and lined them up in a galley.  It was the first repetitive manual manufacturing task and remained essentially unchanged until mechanical typesetting was introduced in 1889.

Printing was really two separate inventions—typography for text and block printing or engraving for images.  Xylographic prints had been common in Europe since the end of the 14^th century, and it was not a great leap to figure out how to use engraved blocks to insert images into printed pages, especially those made perfectly flat by a press. Illustrations first appeared in a printed book only seven years after Gutenberg’s Bible. Religious images were followed by illustrated posters and then playing cards and calendars, and science turned to pictures shortly after that. Jesuit scholar Walter Ong said printing with illustration made quantification of knowledge possible. Exquisite illustrations could be reproduced exactly as they were drawn. Maps and books on zoology, botany, and metallurgy were published in the first hundred years, but perhaps the most stunning was Vesalius’s De humani corporis fabrica libri septem illustrated by Titian’s pupil Jan Steven van Calcar. For the first time physicians and artists could see the interior of the human body as it actually is.

We have neither the plans nor a picture of Gutenberg’s press, but it was probably much like the ones in use for the first three and a half centuries of printing and was directly modeled on wine presses.  The galleys were secured in a hinged wooden frame—the forme—mounted on a cart on rails since there was not room to ink the page under the press.  The forme also had paper flaps around the edges so only the part of the sheet to be printed was exposed to ink and the margins could be kept clean.  Once the forme was closed, the letters were inked and the paper was laid across them.  The dolly with the prepared forme was cranked over a marble bed covered with a layer of felt or sheets of paper that allowed the sorts to give a little in case they were not exactly the same height. The platen was then screwed down on the page.  The whole device was relatively simple and could be reproduced by any skilled carpenter, so presses multiplied across Europe.  The process remained essentially the same until 1814 when the Times of London used the first mechanical press to crank out 1,100 sheets an hour.

If an average print run was 500 copies (a conservative estimate), based on the 30-50,000 editions of books known to have been printed between 1454 and 1501, about 20 million books were produced in the first half century after Gutenberg’s Bible.  In the next hundred years, another 150-200 million were printed on a continent with a population of only 20 million, most of whom could not read.  The economics were revolutionary. In 1483 a print shop in Florence run by nuns from the Convent of San Jacopodi Ripoli sold 1,025 copies of Plato’s Dialogues for three florins. A scribe would have charged one florin for a single hand copied version of the same book.

Those four bits of technology changed how information was transmitted in a stunningly short period.  On March 12, 1455, Eneo Sylvio Piccolomini wrote the future Pope Pius II: “What was written to me about that marvelous man seen at Frankfurt is entirely true.  I have not seen complete Bibles but only a number of quires of various books.  The script is extremely neat and legible, not at all difficult to follow.  You would be able to read it without effort, an indeed without glasses.”[4] Mainz managed to keep the details of Gutenberg’s press a trade secret until the city was sacked in 1462.  Refugee printers scattered the technology to Ulm, Basel, Venice, and Rome.  In a decade, there were presses all over Europe.

We left off earlier without finishing the story of that “marvelous man.”  Gutenberg borrowed 500 florins at 5 per cent interest and then 300 more from Johann Fust to cover the formidable cost of his press, paper, and fonts.  He had made some money printing indulgences and short tracts, but they did not nearly cover his expenses and he was unable to pay back either principal or interest on the money he borrowed.  Fust and the calligrapher Peter Schoeffer took over the press, which Schoeffer successfully ran well into the 16^th century. Gutenberg dropped from view.  He was ennobled by the Elector of Mainz in 1465 but remained poor, and no book with his name on it was ever published.  So it goes with inventors and investors.

As is almost always the case with “inventions”, printing was an ingenious combination of technologies that mostly existed to fill a need almost no one knew they had.  The process was perfected and the product became indispensable.  The way knowledge spread and the way the brain worked abruptly and irrevocably changed.  We are living in a similar time.

 References:

Chappell, Warren, A Short History of the Printed Word. Boston: Nonpareil Books, 1970.

Eisenstein, Elizabeth, The Printing Revolution in Early Modern Europe. Cambridge: Cambridge University Press, 1983.

Febvre, Lucien and Henri-Jean Martin, The Coming of the Book: The Impact of Printing, 1450-1800. London: Verso, 1976.  The standard reference.

McLuhan, Marshall, The Gutenberg Galaxy. Toronto: University of Toronto Press, 1962.   Third Edition 2011.

Raven, James, ed., The Oxford Illustrated History of the Book. Oxford: Oxford University Press, 2020.  The best current summary by multiple authors.

[1] The chips also found a niche in bitcoin mining although that market has been unstable recently.

[2] Wood fiber paper did not become available until the 1840s.

[3] Lead poisoning was a common occupational hazard in print shops.

[4] Martin Davies, trans. “Juan de Carvajal and Early Printing” The Library 18:3 (1996) 196.

                No matter how you define intelligence, memory is its fundamental requirement.  We have talked a bit about the variations in memory’s importance over time—writing that replaced oral history and printing that obviated the need to memorize manuscripts—but we have not looked at how memory actually works.  How is it that fully formed images of something that happened decades ago suddenly pop into our heads?  How were the images made and where in the world (or the head) have they been? That takes us to our first scientist. 

                Eric Kandel was born to a middle-class Ashkenazi Jewish family in Vienna in 1929.  When he was nine years old, he and his older brother were sent to live with an uncle in Brooklyn to escape persecution following Austria’s absorption into Nazi Germany.  In New York he went from a yeshiva to Erasmus Hall High School and then to Harvard where he was determined to find a relation between Freudian psychology and clinical neurology.  Kandel went on to medical school at NYU and then to the Laboratory of Neurophysiology at the National Institutes of Health before completing a residency in psychiatry.  During that time, he started research into the role of the hippocampus in memory, but that proved too complicated for him to unravel.

The roadblock led Kandel to the nervous system of Aplysia californica, the giant sea snail.  The slightly creepy invertebrate had a couple of major advantages. First, it was relatively large and had large neurons.  Second, its nervous system was simple.  It’s gill would retract when the surrounding tissue was touched and the animal could be trained to pull its gill back when its tail was shocked. The activity of the neurons responsible for those reflexes could be measured directly, and the snail got Kandel the Nobel Prize in Physiology or Medicine in 2000.

To understand what Kandel found, we need to start with how a simple reflex works.  The simplest reflex in the human is what happens when a physician taps your knee with a hammer and your leg kicks out.  What happens exactly is that the hammer hits the tendon below your knee cap and stretches it.  That stretch makes a receptor in the tendon fire and an electric current goes up the fibers (dendrites) that go back to the body of the sensory neuron.  From there the message is sent out an axon.  At the end of the axon, there is a knob that lies up against an empty space—the synapse.  When the current arrives, the axon releases a chemical into the synapse and that chemical attaches to a receptor on the next neuron in the chain.  That motor neuron fires when enough of the transmitter chemical is captured and a signal is sent down that cell’s axon to the thigh muscle which contracts.  Out goes the knee. 

This ‘monosynaptic’ reflex is the simplest kind of neural network, and it has two key features that are worth our attention.  First, it is all or nothing.  Either the two neurons fire or they don’t.  Sound familiar?  It is a typical binary system.  The second important feature is that an input (the tendon stretch) causes a signal to be sent to a processor (the sensory neuron) that does something (sends a message to the motor neuron) that does something else (makes a muscle contract).  An input causes a process that leads to an output that becomes the next input where something else is done that causes another output.  We have just described exactly what happens when a computer is programmed—a linked series of binary inputs and outputs. But the brain is much more complex than just a series of a few hundred billion binary, serial processes.  To understand that we need to look more closely at what else Kandel found.[1]

 First, he looked at the difference between short term memory that fades away after a few minutes and long-term memory that can last a lifetime.  Short term memory is a function of the fact that, while the neuron has only one output, it receives many inputs through its dendrites.  Some of those make it more likely to fire and some make it less likely, so the neuron is not actually all or nothing—it is not binary.  Each of the brain’s neurons is an integrator that sums thousands of inputs of varying strength from thousands of sources before sending a signal down its axon. 

Sometimes a long-term memory is laid down.  That happens when input is strong enough to activate an enzyme that causes a section of the neuron’s DNA to uncoil and make a messenger RNA.  That RNA goes back out to the cytoplasm where new proteins are generated that go down the axon and cause new synaptic terminals to grow and remain stable.  That conversion takes place, for the most part, in the hippocampus, but the memories are ultimately stored in the area that generated the initial input—visual memories, auditory memories, emotional memories go back to where they came from and stay there. The important thing to remember is that short term memory is a temporary chemical process and long-term memory involves permanent anatomic changes.

The real magic—and the part no one has worked out yet—comes when the memory is recalled, and all of its various aspects are reintegrated into a whole picture that is the essence of consciousness.  For our purposes, however, there are two aspects of this memory formation that are central to understanding artificial intelligence.  First, this is not a simple binary code.  Whether the neuron fires is a function of a number of inputs of varying strength that are summed.  It is an analog process.  Second, it is not like the monosynaptic reflex in which input and output are in a conga line, one following exactly on what came before.  The brain is a parallel processor in which a large number of inputs of varying strength affect the output. It is far more complicated than the usual computer.  And that brings us to our next fascinating individual.

John von Neuman was born in Budapest in 1903 to an upper class, non-observant Jewish family.  By age eight, he knew eight languages (including ancient Greek) and was adept at both differential and integral calculus.  He published his first major mathematical paper at age nineteen and went on to publish more than 150 others, mostly theoretical and applied mathematics, physics, and computer science.  In 1933 he came to the Institute for Advanced Study at Princeton and later worked on the Manhattan Project.  After World War II, he was responsible for the concept of mutual assured destruction (MAD) to avert nuclear war.  Albert Einstein’s Wikipedia entry runs to thirty-one pages when printed out.  Von Neuman’s is ninety-one. 

In his mid-fifties, von Neuman contracted cancer (the exact type is not clear), but he spent his last months working on a series of lectures on artificial intelligence.  He was not well enough to finish and deliver the guest lectures he had scheduled, but what he had completed was published as The Computer & the Brain with an introduction by futurist Ray Kurzweil.  Von Neuman knew that computers were much faster than the brain and had a much lower error rate.  He died before transistors were developed, but even the vacuum tubes in von Neuman’s ENIAC computer were 100,000 times faster than the simple neural circuit. 

If the brain was a serial processor like the computer, it could not possibly compete, but it isn’t.  Von Neuman played a mind game to estimate the brain’s ability to store information.  He assumed that no long-term memory was ever lost.  He guessed that every neuron was constantly receiving input from its thousands of dendrites and guessed that amounted to an average of fourteen inputs a second to every cell.  Since there are roughly 10¹⁰  neurons that would mean 14 x 10¹⁰  inputs a second.  If one lived an average of 60 years (a fair estimate in the early fifties), that would be 2x 10⁹ seconds x 14 x 10¹⁰  inputs or 28 x 10¹⁹ bits of stored memory—about 100 million of the fairly powerful auxiliary hard drives sitting next to my computer.  It is not just a matter of storage.  To match the brain’s processing power, a digital computer would have to perform an impossible 10¹⁶ operations a second.

But all that ignores the fact that the brain is not a serial processor.  Each neuron ‘processor’ has weighted inputs and can do multiple simultaneous operations.  What it lacks in speed, the brain more than makes up for in complexity. That is about where von Neuman left it when he died in 1955, and that is where it more or less stayed. 

Frank Rosenblatt actually proposed a computer with a parallel processing system and weighted inputs for the IBM 704 computer in 1957, but the existing hardware was not up to the task.  All that changed at the end of 2017 when Ashish Vaswani and his co-authors published “Attention Is All You Need” in which they described the generative pre-trained transformer.  In that model, a computer is trained with a gargantuan set of historic data and asked to predict specific outcomes.  To do that, the machine takes multiple inputs simultaneously, assigns weights to them, and tests its predictions against reality.  It then adjusts the weights until the machine’s prediction matches reality as closely as possible.  In many ways, it is doing precisely what the brain does but with much more information and much faster. Like the brain, the GPT model is a parallel analog processor but with a knowledge base and speed no human can possibly match.

The question is whether there is still something about the human brain that is unique.  That is a question for another time. Meanwhile, generative artificial intelligence looks very much like an inflection point in human history.

If you want more depth:

Kandel, Eric R.,  In Search of Memory: The Emergence of a New Science of Mind. New York: W.W. Norton & Company, 2006.

John von Neuman, The Computer & the Brain. New Haven: Yale University Press, 1958.

Vashwani, Ashish, et.al., “Attention Is All You Need,” arXiv:1706.03762v5, 6 December 2017.  Accessed 3/13/23.

[1] I am going to simplify this a good deal.  To get the details of this fascinating story, try his book In Search of Memory: The Emergence of a New Science of Mind.

The viral response to the November 2022 release of ChatGPT (Generative Pre-Trained Transformer if you care) is the front end of a fundamental change in our world.  As we saw in the last post, Samuel F.B. Morse used his binary code to transmit a message over forty miles of wire in 1844.  Thirty-two years later Alexander Graham Bell sent a human voice over similar wires.  Nineteen years after that (August 1895) Guglielmo Marconi transmitted code through the air, and Kenjiro Takayanaki did the same with pictures in 1926.  But as impressive as that technology was, it was only moving information from human to human and mostly in one direction.1  In 1989 Tim Berners-Lee invented the World Wide Web.  The web allowed people to communicate bidirectionally and also allowed people to query the giant databases resident in machine memory.  Rudimentary artificial intelligence allowed people to ask computers simple questions—“What is the average temperature in Boise in November?”—in natural language. Communication was not just human to human; it was human to machine and the machine answered.

ChatGPT was an accidental offshoot of the large language model that powered Siri and Alexa, but it is something altogether different.  The system has been trained on millions of books and billions of words to be able to pull elements from its memory, see the connections between those elements, and format answers to queries in sentences, paragraphs, or essays.  Conversations are generated that sound as if they were created by a human being.  Chat GPT’s ability to converse in proper grammar in multiple languages while drawing on more data than any human could master in a hundred lifetimes makes it unique.  For now, it is an infant that makes mistakes and has limitations, but conversations between us and machines that know more than we do will inevitably change education, economics, and politics as well as how we see ourselves and our reality.  We are not ready.

If history is more than an entertaining hobby, it is because experience determines how we approach the future.  We use what has happened in the past to build models of what to expect when similar situations arise, and we use those models to plan for what to do when things change.  The closest historical analog to the impact of artificial intelligence is probably the printing press.  Perhaps what happened after Johannes Gutenberg printed his Bible in 1454 can help us predict what might happen with AI.2

AI did not start with Chat GPT, and printing was not new when Gutenberg built his press.  Wood block images coated with ink had been pressed onto paper in China, Japan, and Korea since the seventh or eighth century, but they were pictures not books.  Koreans and Uyghurs had even made blocks of whole words, but that was still not printing as we know it. 

Gutenberg’s invention relied on three things—paper, fat-based ink, and letters.  Paper came to Europe from China in the thirteenth century, but it was not durable enough to justify abandoning papyrus and parchment. A hand-copied manuscript took too much work to risk being easily destroyed, and both parchment and papyrus were stronger than paper, but that kind of durability was not necessary when reproduction was cheap and easy.  Ink had been around for thousands of years, but Gutenberg created an oily compound that worked on the press.

It was letters that made the magic.  Unlike logographic scripts with thousands of characters—Chinese, Japanese, Korean—the European alphabet had only twenty-six or so letters.  A mold of each letter could produce tens of thousands of identical copies.  The ‘sorts’ were stored in pigeon holes in a cabinet at the typesetter’s desk with small letters in the bottom slots and capitals in the ‘upper case’. The more common the letter, the more copies of it in the cabinet—a lot of e’s and only a few q’s.  The printer pulled the letters one at a time and lined them up on a composing stick that was set in a tray or galley and held in place with wooden wedges.3  The galley was inked, covered with paper, and pressed. Out came sheet after identical sheet.

Gutenberg suffered the fate of many innovators; his financiers forced him out. He vanished after 1455 and likely died poor, but his press thrived.  By 1500, every major city in Europe had a printing shop. Eight million books were printed before the end of the century, more than all the volumes laboriously copied by scribes in the previous thousand years.  In the next hundred years it is conservatively estimated that 150-200 million books were printed, far more than there were people able to read them.

The press was, although composed of previously known parts, a new technology.  Bound paper replaced scrolls.  Because the copies were identical in format and pagination, tables of contents and indexes were possible; one no longer had to go through an entire scroll to find a piece of information.  Research was possible, and works on the same subject could be compiled and compared.  

Printed books were the first items to be manufactured in identical copies, and printing made money.  Manuscript writing was a craft and occasionally an art, but  printing was a business that presaged other mass production—cloth from Frances Cabot Lowell’s looms, interchangeable parts for Samuel Colt’s revolvers, and Henry Ford’s Model T’s for example.   

It is hard to overestimate the change wrought by printing.  Marshall McLuhan said, “The difference between the man of print and the man of scribal culture is nearly as great as that between the literate and the non-literate.”  The change permeated every aspect of western society.

For the first two hundred years most printed books were reprints of the classics—77 percent of incunabula (books before 1500) were in Latin, and most of the rest were in Greek.  If the Renaissance began with printing, it first half was taken up with collecting and imitating what had been created a millennium or so earlier.  The second half was, however, different.  Printing freed the brain from the obligation to memorize, and the free time could be devoted to categorizing and analyzing the flood of information in newly available books.  A group of innovators—men like Leonardo da Vinci, Paracelsus, Ambroise Paré, and Andreas Vesalius—created new knowledge and used the press to spread it. The explorers and Nicolaus Copernicus changed how Europeans saw the earth and the skies.  None were Latin scholars and many could not even read the language, but exact copies of what they wrote along with repeatable, detailed maps and illustrations came off the presses and created the Renaissance scientific revolution.

Science was not the only information spread by the printing press.  Destruction of the status quo in religion predated the scientific revolution and, at least at the time, had a greater impact.  Between 1517 and 1520, Martin Luther published thirty texts that sold 300,000 copies and broke the Vatican’s monopoly on Christianity. 

One can argue that the Council of Trent (1545-1563) was just the Catholic Church’s attempt to control the printing press.  Vulgate Latin was declared the official language of the Church, and attempts were made to restrict both vernacular versions of scripture and limit who could read it.  The Sacred Congregation of the Inquisition published its index of prohibited books and began arresting people spreading Copernican doctrine in 1616.  Despite that, presses poured out books in local languages accessible to the general public.  Pride of language drove tribes to become nations, and the competition between new nations and religions were a proximate cause of the brutal warfare that tormented sixteenth and seventeenth century Europe.  In the end, the Church failed to stop printing; the intellectual center of gravity merely shifted from the Italian peninsula to northern Europe.

England’s kings found a more productive way to deal with printing.  Rather than try to abolish it, they used it.  The press spread decrees and propaganda and was a major reason the Tudors were able to enhance their power over nobles, judges, and Parliament. 

So what are the lessons from the printing press that might be applicable to artificial intelligence?  First, a qualitative change in how we communicate will probably spread much more widely and far more rapidly than we expect.  The change is unlikely to be gradual. Second, some intellectual skills that we currently see as important will atrophy and be replaced by new ones. Third, AI will generate knowledge we cannot foresee. That may come in stages with the early efforts being enhancements of what we already have, but, as the technology matures, huge databases and neural networks are likely to generate things we cannot predict.  We will surely get vast amounts of new knowledge, but we may not get the understanding to go with it.  The status quo and social stability are not safe.   Fourth, attempts to stop the technology will almost certainly fail. Societies and governments that best learn to use the new technology will be the most influential, the wealthiest, and the most powerful. 

At least that is what happened after 1454.

FURTHER READING

Eisenstein, Elizabeth, The Printing Revolution in Early Modern Europe. Cambridge: Cambridge University Press, 1983.

Gershberg, Zac and Sean Illing, The Paradox of Democracy: Free Speech, Open Media, and Perilous Persuasion. Chicago: The University of Chicago Press, 2022.

Kaku, Michiko,  The Future of the Mind: The Scientific Quest to Understand, Enhance, and Empower the Mind. New York: Doubleday, 2014.

Kissinger, Henry, Eric Schmitt, Daniel Huttenlocher, “Chat GPT Heralds and Intellectual Revolution. Generative artificial intelligence presents a philosophical and practical challenge on a scale not experienced since the start of the Enlightenment” Wall Street Journal, 2/25/2023.

McLuhan, Marshall, The Gutenberg Galaxy. Toronto: University of Toronto Press, 1962.   Third Edition 2011.

Invention of the “electro magnetic telegraph” says a good deal about how new technologies are born.  First, almost nothing of the first American telegraph was new.  The idea had been worked on for almost a century. Several working models had been built, but none was a practical, reliable way for electricity to move information.  Second, the binary code that would fundamentally change communication was a late development. That code was essential to the invention and would be its greatest legacy, but the man given credit for it did not understand its importance.

Samuel Finley Breese Morse was a difficult man from a difficult family.  His father Jedediah was a Congregationalist pastor and author of geography and travel books with sales that rivalled those of Noah Webster’s spelling books and the Bible in both Britain and the United States.  He socialized with Benjamin Franklin, George Washington, John Adams, and French minister Charles Maurice de Talleyrand-Périgord and set a standard his oldest son spent a lifetime struggling to match. 

Finley—he was never called Samuel—had a younger brother Richard who suffered repeated bouts of depression and died relatively young of liver cancer.  Morse’s oldest daughter Susan also had incapacitating depression, and her son Charles committed suicide.  His namesake son was mentally incapacitated and was never able to live independently. Willie, a son from his second marriage, was briefly incarcerated for murder before joining Buffalo Bill’s Wild West Show.  His son Arthur, also from the second marriage, either fell or jumped from a railroad platform and was crushed beneath a moving train.

Morse also suffered from repeated episodes of depression and seems never to have had a close friend.  His business partners uniformly disliked him and several sued him.

Both Finley and his two brothers went to Yale where they thrived while he barely made it through.  Although he was a poor student Morse could paint, and he earned pocket money selling portraits.  He wanted to take up painting as a profession, but Jedediah insisted he apprentice at a bookstore.  That lasted just three months before the younger Morse convinced his father to send him to London for three years to work with prominent American artist Washington Allston.

  Morse’s father scraped together an $800 annual allowance and increased it to $1000 when the young man complained he could not afford fruit, new clothes, or even “nice dinners”.  Jedediah offered to let him copyright his new American geography in London and keep the profits, but the aspiring artist considered the boredom of business and accounts beneath him.  Even though it required going into debt, Morse’s father was badgered into funding still another year in London before Finley agreed to come home provided painting would be his only job.  He wanted to create great historical paintings, but there was no  market for those so he was forced to become an itinerant portrait painter at $15 a commission.

Besides art, Finley was open to other ways to become wealthy.  He and his brother Sidney patented a water pump and promoted it for use in fire engines, but, as with the London copyrights, Morse had no interest in the mechanics of starting a business, so the invention went nowhere.  He enrolled in an Episcopal seminary but only lasted there two months before going back to portrait painting.  This time he went to Charleston where wealthy planters were willing to pay as much as $70 for one of his paintings, and he was briefly solvent for the first time.

Morse went back to New Hampshire in 1817 and married Lucretia Walker whom he called Lucrece.  A worsening South Carolina economy and growing dissatisfaction with his generally unflattering portraits forced Morse back to New England where another try at grand historical art failed.  In 1824 he took another shot at entrepreneurship with an attempted patent on a marble carving device.  That too failed.  He next tried to get an appointment as secretary to the newly appointed American ambassador to Mexico—another failure.  He did win a commission for a commemorative portrait of the Marquis de Lafayette, but when it was done critics were generally unimpressed.

While he was in New York working on the Lafayette portrait, Lucrece died giving birth to their third child.  He left the children with relatives and stayed in New York where he was elected an Associate of the American Academy of Fine Arts. He promptly got in a dispute with the American Academy and started a rival National Academy of the Arts of Design.

In 1826, Finley’s mother wrote asking him to come home for a final visit with his rapidly dying father.  He wrote back that he was too busy organizing an exhibit at his academy.  Jedidiah died leaving his wife with a substantial debt, three unsuccessful sons, and Finley’s three children.

In 1830, Morse went back to Europe where he supported himself doing portraits and making copies of old masters. He stayed until 1832 before taking the steamship Sully from Le Havre for New York, a voyage that changed his life.

On returning to New York, Morse was named Professor of the Arts of Design at the new University of the City of New York (eventually renamed New York University).  The position carried no pay except fees he was expected to collect from his students, but he did take the title Professor that he used the rest of his life. He also became vociferously anti-Catholic and anti-immigrant.

Morse had spent more than two decades unsuccessfully trying to use art to become as famous as his father,. It was time to change course, so he set to work on an idea he said came to him on the Sully.

Charles Jackson, a twenty-eight-year-old Harvard trained physician, was on the Sully as well, and he and Morse spent their idle hours discussing how electricity might carry information.  The idea was far from original.  Three quarters of a century earlier, a Scotsman—probably Charles Morrison—anonymously published the idea of using static electricity from a Leyden jar sent through thirty-six wires each of which, when charged, could pick up a piece of paper with a letter printed on it.  Charge the right wire and a letter appeared on the other end.  It was never built, and the lack of adequate insulation would probably have prevented its working anyway, but a number of inventors in England and Europe started working on other ways to communicate with electricity. It is likely that Morse did not know about those efforts.

 At any rate, when he got back to New York, Morse started building a model “telegraph” in his rooms at the university, an effort severely hampered by his lack of funds and a serious lack of mechanical ability.

It is worth going through his device in order to see which of the parts were original.  There had to be an electric current and wires to carry the current from the battery to a receiver.  He needed a mechanism to open and close the circuit since whether the current was on or off conveyed the information.  There also had to be a way to receive the signal, and Morse, unlike the Europeans, wanted to make a record of the signal.

The battery was a simple voltaic cell, and the wires were copper. Neither was new.  At the transmitting end, the wire was broken and each end could be dipped into one of two small containers of mercury joined by a wire.  When the ends were in the air, the circuit was broken.  When they dipped into the mercury, it was complete.  The wire yoke was attached to a lever the other end of which had a metal ridge.  Beneath that end of the lever was a notched stick that could be moved so the protruding parts moved the lever up and down.  Pulses of electricity with a pattern determined by the notches were sent down the circuit. This “port-rule” was original.

At the receiving end, the pulses of electricity had to be turned into a mechanical force to create a record.  That relied on the well-known fact that wrapping a wire around a soft iron bar turned it into a magnet when the wire carried a current.  The magnetized bar swung a pendulum with a pencil on the other end.  Beneath the pencil was a roll of paper tape moved by a clockwork mechanism.  When the current flowed, the magnet moved the pendulum and pushed the pencil from side to side. The straight line on the paper jerked into a V.  The series of V’s on the paper exactly mirrored the opening and closing of the circuit caused by the pattern of notches on the port-rule dragging across the lever.  It took Morse five years to build this Rube Goldberg apparatus.

The final necessity—and the key one—was some way of turning the notches on the port-rule into words.  It could generate numbers—for instance vv vvvv v is 241—but numbers were not words.  Morse’s solution was a dictionary that grew to almost 30,000 entries in which each word was given a number.  The dictionary took a great deal of work, and anyone using the telegraph would have to either memorize the numbers or look them  up.

There were other problems.  After five years work, the device would only  carry a current thirty of forty feet before it degraded and the message was lost.  Morse showed his telegraph to Leonard D. Gale, professor of chemistry at the university, who solved the problem.  First he gave Morse a more powerful battery.  Then he showed him how wrapping the iron bar more tightly made it a more efficient magnet.  Morse’s tens of feet became hundreds, and, in 1837, he was able to mount a demonstration for a visiting Oxford professor using 1,700 feet of copper wire looped around the periphery of a large room at the university.

The demonstration was seen by Alfred Vail, one of Morse’s students who just happened by.  Vail was captivated, and his enthusiasm changed the telegraph.  The young man’s family owned the Speedwell Iron Works in Morristown, New Jersey. The foundry had built the shaft of the first steamship to cross the Atlantic and was turning out wheels, axles, and cranks for the rapidly growing United States railroad industry.  Alfred’s father, Judge Steven Vail, had money, and the foundry had a machine shop.

In February 1837, a United States House of Representatives resolution requested that the Secretary of the Treasury look into a “system of telegraphs for the United States.”  Here at last was a chance for Morse to become rich and famous if Vail and his family could help him build a good enough model to convince the powers in Washington he had the solution.  Vail agreed to supply $2000 and his own mechanical expertise in return of one-fourth of the rights to the invention in the United States and half interest in any foreign patents he could obtain.

Vail, assisted by fifteen-year-old apprentice William Baxter, retired to a locked room at Speedwell and set to work.  He used Gale’s battery.  There was no source of insulated wire so he and Baxter bought all the copper milliner’s wire they could find.  Even though the wire was popular for “skyscraper bonnets” and plentiful, they depleted the entire New York garment district supply and hand-wrapped it with cotton thread.  Vail next fixed Morse’s register.  He traded the pencil that required repeated sharpening with a fountain pen and later with a stylus that made dents in the paper rather than marks.  He also made the pendulum swing in the same direction as the paper moved so the V’s turned into dots, dashes, and spaces. 

Several days of demonstrations for President Martin Van Buren, his cabinet, and members of Congress began February 15. The test was a success; the House of Representatives Committee on Commerce asked for a full report on the invention and recommended a $30,000 appropriation for a fifty-mile trial line.  Committee Chairman Francis O.J. Smith was so impressed he left Congress to become a partner in Morse’s venture. Unfortunately for Morse, Vail, and Smith the wheels of Congress grind slowly. It was 1843 before the money was allocated to build a line from Baltimore to Washington.

During those six years, Vail was not idle.  The code was an issue. Morse had put an inordinate amount of time into his number-based dictionary, and he was loath to give it up.  Who actually created the code that carries Morse’s name is a point of contention.  In support of Morse’s claim biographer Kenneth Silverman cites a note from Vail to his father dated February 7, 1838: “Professor Morse has invented a new plan of an alphabet and has thrown aside the dictionaries.”  But Vail went on to say Morse’s alternative was a “very imperfect” combination of dots and lines of different length, and, when a demonstration of it failed, Morse reverted to the number dictionary for the Washington demonstration. In a letter to Vail’s son, a former acquaintance said Alfred told him that the code “came to him like a flash” one day when he was in the offices of the New York Observer.  Vail later wrote, “Soon after my connection with Professor Morse as copartner, and at the time I was constructing an instrument for exhibiting the advantages of his telegraphy to a committee of Congress, it occurred to me that a plan might be devised, by means of which letters of the alphabet could be employed in recording telegraphic messages.”  His assistant supported that; as soon as Vail invented the register that could make dots and dashes instead of V’s, Baxter said he “saw in these new characters the element of an alphabetical code by which language could be telegraphically transmitted in actual words and sentences, and he instantly set himself at work to construct such a code.”

According to Baxter. Vail planned to have the most common letters have the shortest code; e was . and T _.  Less used letters could have longer codes; Q was__._ for instance.  Baxter said that, upon returning to Morristown, Vail went to the offices of the Democratic Banner and counted the number of letters in the typesetters’ cases assuming that the most common characters would be the ones most used and needing the shortest code.

Why did Vail not claim the code if he had invented it?  In the first place, he never publicly claimed any of his improvements, assuming his contract with Morse prohibited his doing so.  In addition, Morse’s attorney Amos Kendall warned Vail that any patent dispute might invalidate his contract and take away his interest in the company.

Regardless, Vail also invented a telegraphic key to replace the port-rule.  That combined with the binary code made electrical transmission of messages practical.  There was a wire from London to Paris in 1852.  By 1855 there were 42,000 miles of wires in the United States alone.  In August 1858, after a plethora of difficulties, the first trans-Atlantic message was sent by Queen Victoria to President James Buchanan.  It took a few more years for a reliable cable to be laid, but it lasted until it was replaced by fiber optic cable in 1966.  The United States military did not phase out Morse Code until the 1990’s, and the last ship-to-shore message in the code was sent in 1999.

Morse had done what good inventors do.  Very little in his telegraph was really original;  almost all the parts had been tried before.  That said, with the help of Vail and others, he managed to pull the parts together and revolutionize communication.  Printing had removed space as a limitation on the spread of information; the telegraph removed time. 

It is unlikely that either Morse or Vail recognized it explicitly, but the pivotal element in electronic communication was the first binary code that allowed a simple on/off in a circuit to be converted into intelligence and recorded. By the late nineteenth century, wireless transmission and analog wave-based signals pushed binary coding aside, but that was temporary.  Silicon chips and ASCII coding changed dots and dashes into 1’s and 0’s and binary code came roaring back.

Share The Changing Brain

Leave a comment

REFERENCES

Marland, E.A., Early Electrical Communication. London: Abelard-Schuman Ltd., 1964.

Pope, Franklin Leonard, “The American Inventors of the Telegraph with Special Reference to the Services of Alfred Vail” Century Illustrated Monthly Magazine  April 1888 from Making of America, vol 35, 924-945.  Ithaca, NY: Cornell University.  Acc. 12/14/22.

Silverman, Kenneth, Lightning Man: The Accursed Life of Samuel F.B. Morse. New York: Alfred A. Knopf, 2003.

Vail, Alfred, The American electro-magnetic telegraph: with the reports of Congress and a description of all the telegraphs known, employing electricity or galvanism. Philadelphia: Lea & Blanchard, 1847.

Vail, Alfred, Early History of the Electro-Magnetic Telegraph: From Letters and Journals of Alfred Vail. New York: Hine Brothers, 1914.

Wheen, Andrew, Dot-Dash to Dot.com: How Modern Telecommunications Evolved from the Telegraph to the Internet. New York: Springer Science+Business Media, LLC, 2011.

We started this series with what literacy did to the brain, but we have not talked about what made writing possible.  Homo sapiens and our predecessors have been around for something like a million and a half years, but, as far as we have been able to determine, humans have only been recording with marks for about 70,000 of those.  The oldest known drawing that clearly represents an object is only 45,500 years old and was found on the wall of a cave in Sulawesi in the Sunda Islands of Indonesia.    

Sulawesi Cave Painting

The gap between the indistinct picture of a warty pig and the Iliad was considerable and took several steps to cross. The first stage, pictograms such as the cave paintings, were symbols depicting objects. Ideograms, the next step up, were symbols that stood for objects or concepts but did not necessarily try to imitate what they represented.  Hieroglyphics, logograms or pictures that stood for specific words, were next.  The next advance was stunningly inventive.  In syllabary writing, the symbol is meant not to stand for an object or a word but for a sound—in this case a syllable.  The final step was a set of symbols that could uniquely stand for any sound in a language.  That step is more difficult than it might seem; there have been thousands of human languages, but only 106 have developed a system of writing good enough to produce literature.                                             

Clay tokens representing objects appeared in the Near East around 8500 BCE.  Not surprisingly that coincides with the development of agriculture since the markers mostly kept track of goods and property.  The tokens were found alongside hollow tennis ball-sized spheres that served as purses so the markers could be moved from ball to ball to keep track of transfers. 

Cuneiform Token

By around 3300 BCE, tokens had evolved into pictograms pressed into clay tablets but they were still used mostly to catalogue and count.  The tablets were too valuable to waste.  They were hardened in the sun rather than being fired so they could be washed, scraped, and reused.  Of the cuneiform examples that survive, about three out of four are administrative lists that do not reflect actual speech—“cows 5  donkeys 14  chairs 8  tables 2.”

Hieroglyphics

Egyptian hieroglyphics came about three centuries later and, since they were better representations of actual speech, served as records of events and not just lists of objects.  Both Mesopotamian cuneiform and Egyptian hieroglyphics were dauntingly complicated with more than 600 characters each, many of which had multiple meanings. That meant writing was restricted to specialists, usually government or religious officials.  Hammurabi (1792-1750 BCE) was the first ruler to use written records to keep up with not only the business of government but also its rules.  Over a thousand years went by before a Babylonian ruler, Ashurbanipal (668-626 BCE), accumulated his great library and laid claim to being the first king able to read and write.  

                      Linear B

Meanwhile, a quiet revolution was taking place on the island of Crete.  Around 1800 BCE Minoan Greeks developed their own writing system independent of cuneiform and hieroglyphics.  Linear A (called linear because lines were made using a stylus to inscribe a clay surface rather than pressing an image into it with a wedge) has not been deciphered, but its successor, Linear B, has been and what it reveals is fascinating.  The language has more than 100 ideograms, but it also has 87 characters that represent sounds.  The characters stand for syllables rather than pure sounds, but the use of phonemes had stunning potential.  If you think about it for a minute, a pictogram or an ideogram is just a reminder of something you already know.  It is a memory aid.  A logogram can introduce a whole new idea.  Linear B might have changed the course of western civilization save for one thing.  It was so complicated that it was only used in the royal court and a few temples.  The system was too difficult for general use, so, when the kingdom fell and the court and temples were destroyed, the language was lost.  For the next four hundred years, the Greeks were illiterate.           

        The Phoenician Alphabet                                      

As Linear B disappeared, a new system of writing emerged in the Middle East.  The oldest traces of Semitic writing were found on rock walls in Wadi el-Hol (Valley of Terror) on the west bank of the Nile near Luxor.  Although discovered in Egypt, the letters are probably related to Canaanite script that originated on the Mediterranean coast of what is now Syria. By around 1250 BCE Phoenicians had completely abandoned ideograms in favor of logograms—symbols that represented sounds rather than things—but  their twenty-two symbols lacked vowels.  A consonant has no sound of its own until it is paired with one of the ways breath can be pushed out of the mouth.  K is nothing until it is joined to a vowel—ka, ke, ki, ko, ku.  The Phoenician alphabet was ambiguous and only an expert who knew the context of the document in which it appeared could read it, and even he might not translate what was written the way the writer intended.  Nevertheless, all alphabets—Hebrew, Greek, Roman, Cyrillic, Arabic, and even Korean—derive from that Semitic original.

Then a breathtaking advance came from Greece, and it is the kind of invention that quite possibly came from the mind of one man.  It is mistakenly claimed that the Greeks merely tweaked the Phoenician alphabet, and it is true that many of the names of Greek letters and even their shapes mimic the Semitic model. But, as we have seen, symbols in the Semitic alphabet stood for syllables and were complex and ambiguous.  The Greek alphabet was a stunning improvement.  It included vowels, those non-sounds that, when added to consonants, make speech.  With the combination of vowels and consonants, the alphabet turned combinations of letters into a code that forced the reader to take in exactly what the writer intended.  One no longer had to be a scholar to be literate; ceramic images dating to 490 BCE even show Greek children learning to read and write.      

Nestor’s Cup

One of the older examples of the Greek alphabet was found in a tomb on Ischia, an island in the Bay of Naples and dates to around 720 BCE.  The inscription reads (right to left as was the custom in early Greek) “I am the cup of Nestor good for drinking.  Whoever drinks from this cup desire for beautiful crowned Aphrodite will seize him instantly.”  I cannot help but wonder what Nestor had in the cup.

Humans can make a wide range of sounds, but only about forty-five of those are needed to make a language.  For an alphabet to work, it must include enough symbols to code every phoneme in its language with a maximum of twenty to thirty shapes each of which has a unique, fixed function.  The letters can either be used alone or be combined to make the requisite number of sounds  such as ou, ch, tr, or pl. 

As with any seismic change, writing was controversial.  By the time Plato was born in 427 BCE, the Greek alphabet was ubiquitous, and the average Athenian could read and write.  Education was no longer restricted to a small elite; a literate student might even know things his teacher did not.  Professional scholars and philosophers were teaching anyone capable of paying for an education, and Plato was certain that was a mistake.  Common people could not be trusted to differentiate truth from opinions formed from unsupervised reading.

 In the end, Plato lost that argument, and Greece became a global power and Greek became the first global language.  After Alexander, Greek was spoken from Italy to Afghanistan;  Babylonians knew Homer, and Sophocles was read in what is now Pakistan and Iran. That dominance was not a matter of genetics.  It was the direct result of the inspired invention of the alphabet.

Before we go, we need a word about Asia where a large part of humanity was ruled by cultures that did not abandon ideograms, those symbols that represent not sounds but whole words. Minimal literacy in Chinese requires memorizing at least 3,000 characters, and mastery necessitates almost 50,000.  Historically, only 1-2 percent of Chinese were truly literate.  Alphabetic script had been known to the Chinese for centuries, but until recently there was a conscious decision not to use it for reasons that Plato would have recognized. Reading by common people was thought so dangerous that, around 213 BCE, the emperor Shihuangdi had every manuscript in his kingdom that did not deal with medicine, agriculture, or prophecy burned.

It has been claimed that the Chinese mastered printing before Gutenberg.  It is true they had wood blocks capable of reproducing characters, but they generated images, not letters.  Without an alphabet, mass production of printed matter never happened. 

Meanwhile in the west the alphabet changed how societies operated, how knowledge was created, and how the brain worked.  But the Greek alphabet may have a shelf life.  In posts to come we will look at the possible effects of changing “I am happy to see you” to “I’m 😊 2 C U”.

REFERENCES  (should you want more detail)

Ong, Walter, Orality and Literacy. New York: Routledge, 1982.  Ong is the recognized expert on the evolution from oral to literate thinking and this book has an excellent discussion of the development of written language.

Havelock, Eric, The Literate Revolution in Greece and Its Cultural Consequences. Princeton:  Princeton University Press, 1982.  Perhaps the best discussion of the evolution from Linear B to the Greek alphabet.

Vallejo, Irene, Papyrus: The Invention of Books in the Ancient World. Translated by Charlotte Wittle. New York: Alfred A. Knopf, 2022.  In spite of occasional factual errors and personal asides, this is a quite readable account.  The prose is delightful even in translation from the original Spanish.

I appreciate you all taking the time to read these posts.  Please feel free to comment and to tap the share button and pass this along to whomever you wish.  The next entry will come after the New Year.  Have a great holiday.

By the sixth century, the western Empire had succumbed to infection, indolence, and invasion.  The thousand years between the fall of Rome and the rise of printing was an age of memory, not nostalgia—although there was plenty of that—but actual memorization.  Literacy was defined as the ability to read and write Latin, and educated Europeans devoted their careers to finding what survived of Greek and Roman literature and committing a staggering amount of it to memory. 

There were a handful of large libraries such as those at Alexandria and Pergamum and scattered collections of a few hundred books mostly housed in monasteries and cathedrals.  Monks spent their days in drafty scriptoria perched on straight backed chairs behind high desks.  Silent reading was a rare skill, so the halls had a constant low hum from scribes mumbling what they were copying.  Reproducing texts was time consuming and expensive, and the copies were shot through with errors, some of which were accidental and some of which were intentional edits by copyists with their own thoughts about what the book ought to say.  A scholar  lucky enough to find an uncorrupted copy of an ancient manuscript had to either travel to the book or pay a healthy fee to purchase new parchment and have a copy made.  The search for books was the core of intellectual life; requests for loans and contracts for copies comprise much of what survives of medieval correspondence.

Scarcity, expense, and corruption of the manuscripts made memory both more convenient and more credible than manuscripts. Through the Middle Ages, the definition of intelligence was having a great memory; imagination was generally mistrusted since it was generally understood that all great ideas originated before the fall of Rome. The amount of classic and religious writing that lived in one’s head not only equated to intelligence but also measured character, judgement, and piety. 

The seven liberal arts were adapted from the ancients as the basis of medieval European education.  The trivium—dialectic, grammar, and rhetoric—were the basic skills that prepared the student for the quadrivium—arithmetic, geometry, astronomy, and music. Rhetoric, the most important of the trivium, was based on memory since a skilled orator was expected to speak for several hours without notes. For medieval intellectuals writing was only a way to trigger memory, and they were prodigious memorizers. 

 Memorization was not a new skill in the Middle Ages.  Seneca asked each student in a 200-member class to recite a line of poetry and repeated every line back in reverse order. Themistocles was said to have known the names of all 20,000 of his fellow Athenians.  Cyrus claimed to know the name of every man in his army, and Mithridates of Pontus knew all the twenty-two languages spoken by his subjects. 

Memory, however, became much more important after 500 BCE. Memorized sagas were the news media of the Middle Ages. Great houses had resident minstrels who wrote poems extolling their personal and family virtues and great deeds they had done.  Good poems were sold to jongleurs who traveled around the country reciting epics that regularly ran to 60-80,000 verses the bards had memorized after only hearing them two or three times.

Religious tracts were also prime targets of medieval memory.  Jewish scholars were expected to memorize all 5,422 pages of the Babylonian Talmud, and it was claimed that, when a pin was passed through one of the sixty-three tractates of the holy book, a scholar knew which letter the pin punctured on each page.  Thomas Aquinas was said to have remembered everything he ever read and much of what he had heard.  It is likely the entire Summa Theologica was dictated to three or four secretaries simultaneously from the “treasure house” of his memory.  Peter of Ravenna knew 20,000 legal extracts, 1,000 texts from Ovid, 200 texts from Cicero, 300 sayings from ancient philosophers, and 7,000 texts from Scripture.

And that brings up the next question.  Did these giants of memory have a brain function that we have lost, was it a matter of practice, or was it just a trick.

First, let’s look at a few tricks.  The most famous dates to Simonides of Ceos who lived from about 556 to 468 BCE and was reputed to be the first poet talented enough to make a living selling his work. He was hired by the Thessalonian noble Scopus to write a poem in his honor.  During the performance, Simonides was called out of the banquet hall, and, while he was outside, the hall’s roof collapsed.  The banquet table and guests were buried in a mass of rubble.  The bodies were mangled beyond recognition, and no one had an accurate list of who had been present.  Simonides realized he could clearly picture everyone there, each in his seat. The room was a vivid, detailed picture in his mind. The “memory palace” technique attributed to Simonides was written down in Ad Herennium between 86 and 82 BCE and became the mainstay of medieval memorization; it was the way Thomas Aquinas, who learned it from his teacher Albertus Magnus, was able to remember so much. 

You can do it as well.  Try picturing your home.  Walk up the drive to your front door.  Open the door and look around the foyer.  Turn into the living room and look around.  Mentally walk through the entire house and stop in each room.  Look at every object and at the details of each decoration, each piece of furniture, every floor and window covering.  By the time you are done, there will be a list hundreds of items long.  Now go back and attach something you want to remember to each object. You will have a much longer list than you could have possibly created by rote memorization. 

Medieval savants did not stop with just one house.  Peter of Ravenna was said to have over 100,000 memory palaces each room of which translated into an exact copy of a text. In the sixteenth century, Giulio Camillo proposed building a real memory palace (a seven-story theater actually) for Francis I of France that would contain reminders of everything ever written or thought—an architectural Wikipedia.  The money to build it never materialized and the plan remained just a plan, but the mental technique remained pervasive.

Scholars realized was that the objects in a memory palace were easier to bring to mind if they were unusual.  Putting a priest in a skirt with a monkey on the living room sofa made it easier to remember what the image stood for.  The easiest cues to remember were salacious, especially images involving naked young women.  Those images were not viewed favorably by straight laced Protestants, so they tried different tricks.

One of the more common ways to memorize text was to visualize the page on which it was written.  Older manuscripts were strings of capital letters without punctuation or even division into words.  A scroll was just a string of letters on a continuous roll that was virtually impossible to visualize, so the letters were divided into pages, chapters, and verses, often with the first words in a paragraph colored or illuminated by a small painting. Note that each of those takes advantage of visual memory. Practiced scholars could picture each page and remember the text it contained.

Poets dating back at least to Homer knew that memory is enhanced by rhythm and rhyme.  You probably learned the alphabet in singsong, and you almost certainly know more song lyrics than Bible verses. That explains why bards recited their sagas in verse and Shakespeare wrote in iambic pentameter.

Printing came to Europe in 1454, and for the first century books about memory techniques were quite popular.  When books became easy to get, scholars realized they no longer needed to memorize them.  Intelligence became less a matter of memory and more a matter of organizing and rearranging what was available.  Except  for students cramming for exams, memory aids have mostly become parlor tricks.

One group that has maintained an interest in those tricks are memory athletes.  Competitive memorizer Yip Swee Chooi memorized all 756,000 definitions in a 1,774-page Chinese-English dictionary.  Ben Pridmore can memorize a randomly shuffled deck of cards in thirty-two seconds and knows the first 50,000 digits of π.  The record for memorizing cards in an hour is twenty-seven decks.  Most of the memory athletes rely on either memory palaces or hierarchies in which specific number combinations stand for words or objects, so at least some of the medieval talent for memory relies on tricks, but not all of it. One of the greatest memory tricks is done by every toddler who learns an entire language and sometimes more than one in a year or two.

Where does memory live in the brain?  In the first place (a figure of speech derived from ‘the first place’ in a memory palace) we do not know exactly how memories are stored.  Is it in neurons or some other brain cells? Is it from new cells or a change in how existing cells are physically connected? Or is it a change in the structure or chemistry within the cells?  What happens in the more than 100 billion neurons each of which have as many as 10,000 connections? We do not know the details, but we do know a bit about where memories are formed.  Canadian neurosurgeon Wilder Penfield stimulated specific areas of the brain in awake patients while he was removing tissue to control seizures.  He found that he could elicit long-forgotten memories by stimulating parts of the temporal lobes and that stimulating the same area again brought back the same memory.  William Scoville made a more unfortunate unintentional experiment when he removed both temporal lobes in another seizure patient.  The seizures were controlled, but the patient completely lost the ability to form new memories.  He could remember a great deal of what happened before the surgery, but every day, every person, and every experience from that day forward was entirely new.  People he had met and places he had been just hours earlier completely vanished from his mind.

At the other extreme, there are also occasional people with remarkable ability to remember—think Kit Peek, the model for the movie Rain Man.  The interesting thing about these “savants” is that virtually all have damaged brains, and that damage is most often in the left side of the brain.  Allen Snyder used transcranial magnetic stimulation to temporarily suppress the dominant hemisphere in otherwise normal humans and found that some temporarily exhibited artistic or musical skill they had previously lacked.  Perhaps a flood of information into and out of the dominant hemisphere from some of the 10 billion books printed every year and billions of digital messages we receive affects our ability to remember.  Printing may have robbed us of the ability to learn a new language the way a two-year-old can or to memorize thousands of pages of text like Peter of Ravenna, and digital communication has almost certainly made the situation worse.

There may be a real cost to the decline and denigration of memory. Modern educators do not look kindly on memorizing.  E.D. Hirsch said corporal punishment is viewed as the only thing more antithetical to education than memorization.  We promote “critical thinking” as the cornerstone of liberal education, but Hirsch argues that critical thinking is impossible without a knowledge base.  There are certain things one must memorize before one can reason.

 Memory plays a fundamental role in the generation of ideas.  Concepts and inventions virtually never spring up de novo. Creativity is the ability to take disparate bits of information from memory and combine them into something new—to make new connections between old ideas.  The more bits we carry in our brains, the more likely it is that we will find creative connections.  The idea that memory is unnecessary in the age of Google may not only be false; it may be very costly.  It remains to be seen whether the generations brought up with information in their devices rather than in their memory are more or less productive and creative than those who have internalized that information.

Printed books changed our brains.  Some of the changes were for the better. Some may not have been.  The same can be said for writing and for the new digital information explosion, and we will continue to examine that.

REFERENCES (A short list of useful additional reading)

Carruthers, Mary. The Book of Memory: A Study of Memory in Medieval Culture. Cambridge: Cambridge University Press, 1990.  An excellent summary of the role of memory in the intellectual life of the Middle Ages.

Foer, Joshua.  Moonwalking with Einstein: The Art and Science of Remembering Everything. New York: The Penguin Press, 2011. A very readable book about the memory athletes with a nice summary of the role of memory before printing.

Hirsch, E.D., Cultural Literacy: What Every American  Needs to Know. Boston: Houghton Mifflin, 1987.  A somewhat controversial condemnation of the denigration of memory in modern education.  The book includes an appendix of items and topics that Hirsch considers essential information to have in one’s memory.  See how many you know and try the list on a twenty-year-old.

Yates, Frances. The Art of Memory. London: The Bodley Head, 2014.  Dense and academic but probably the most complete discussion of the various memory aids used in the Middle Ages and early Enlightenment.

Rembrandt’s Aristotle Contemplating a Bust of Homer. These two philosophers bridged the change in brain function that coincided with Greek literacy.

In the introduction to this series, I suggested that speech, writing, printing and coding each qualitatively changed how brains work.  I also said we would not necessarily  look at them in chronological order, so I will start with the book that started me thinking about brain changes caused by those inventions.

 Julian Jaynes was a curious, controversial figure who earned his bachelor’s degree at McGill and started graduate school at the University of Toronto before spending three years in a federal penitentiary for refusing to join the military or participate in the conscientious objector program during World War II. After the war, he finished his doctoral studies at Yale but refused to submit his dissertation in a dispute with the faculty and administration. He spent the next  fifteen years in London acting and writing plays while taking occasional  adjunct teaching jobs. 

Jaynes eventually returned to Yale and then went to Princeton as an instructor and research associate. He never sought tenure and, although he was a popular instructor and guest lecturer, he stayed at the bottom of the academic ladder until he retired.  He lived a mostly solitary life plagued by alcohol abuse and died following a stroke at age 77.

Jaynes’s obsession with defining consciousness led to his only book—ponderously titled The  Origin of Consciousness in  the Breakdown of the Bicameral Mind—in which he makes the case that the brain fundamentally changed when civilization became literate.  He based the argument on several ideas, some of which are widely accepted and many of which are not.

First, the two cerebral hemispheres do not do the same thing and are not equally important, especially in communication.  The left hemisphere almost always controls the right side of the body and the right hemisphere the left.  More significantly, damage to specific areas in the left hemisphere lead to loss of the ability to understand and create speech and to read and write.  When humans became literate, the left side became vastly more important, and Jaynes believed that literacy made the left hemisphere dominant.  The right hemisphere seems to be more involved in visual-spatial relations and may have a role in some types of creativity, but one can lose that side of the brain and still communicate quite well.

Jaynes argues that something about how the brain processes input and is able to think about itself changed at about the time alphabetic writing was invented.  His ideas became controversial with the suggestion that, prior to literacy, people did not reason out their actions or make conscious decisions.  Instead, in a time when the two hemispheres were equally important, people heard and saw hallucinations from their right hemispheres that seemed entirely real and they obeyed the voices that came from their own brains. Pre-literate people believed the voices and visions were supernatural. They believed they were gods.

Jaynes thought ancient literature proved that hypothesis.  He argued that the earliest writing (other than cuneiform and hieroglyphics that were basically used for counting and cataloguing) were transcriptions of stories that had been passed down orally from the time before literacy.  Those stories, according to Jaynes, preserved the way pre-literate civilizations thought and hallucinated.

The Iliad is the oldest written document for which we have a translation accurate enough get a glimpse into the author’s mind.  It was likely written in the 8th century BCE and was taken from poems aoidoi (poets and bards) passed down after the fall of Mycenae four hundred years earlier.  That places the Iliad within just a few centuries of the first writing and makes Homer a window into the pre-literate brain. If that is true, what we see through the window is fascinating.

 Jaynes argues (with convincing examples) that the characters in the Iliad  never sat back and considered whether they should  eat, drink, make love, or fight.  A rational mind did not enter into the decision.  In fact, there is no word for mind anywhere in the Iliad.  The Greeks and Trojans simply did what the gods they lived with told them to do.  They followed an inner voice or thumos which Jaynes calls hallucinations. Pre-literate Greeks heard and saw gods that were entirely real to them, and they lived their lives obeying orders from those gods.  Without literacy, the left brain was not in control; the right brain, which generated the hallucinations, was equally influential.

 By the 6th century BCE, the Greeks had an  alphabet and were largely literate.  The thumos had been supplanted by the noos or conscious mind and the gods had left the room. Aristotle, Thales, Anaximander, and Pythagoras were exploring how nature worked without the gods being physically present.

The Iliad was not the only written document that seems to bridge the non-literate with the literate brain.  The Old Testament has examples of the same phenomenon. The Book of Amos, written about 760-750 BCE, is the oldest  of the Bible’s prophetic works.  It documents encounters between God and a shepherd for whom the Deity was clearly a physical reality.  By the  8th century BCE, most Hebrews had lost their incarnate God, but Amos and a few other prophets still heard and saw Him. 

Skip ahead to Ecclesiastes which comprises lessons from a philosopher (traditionally said to have been Solomon) that might have been written as late as the mid-2nd century BCE.  Although the book says a great deal about living a life acceptable to God, the Deity is no longer physically present.

Psalms is interesting.  It is a series of prayers composed over several centuries, many of which complain that God was longer present.  Psalm 22 is typical.  “My God, my God why hast thou forsaken me?  Why art thou so far from helping me, from the words of my groaning? O my God, I cry by day, but thou dost not answer . . .”  The angst of having lost a physically present God was not unique to the Hebrews.  The Assyrian Tukulti-Ninurta from approximately 1230 BCE is strikingly similar. “My god has forsaken me and disappeared.  My goddess failed and keeps at a distance.  The good angel who walked beside me has departed.”  By the beginning of the first millennium BCE, God or the gods had gone to the heavens.  By around 400 BCE, even the prophets no longer saw or heard them. All that remained were the sacred texts.

Jaynes argued that the loss of  the gods was really the loss of visual and auditory hallucinations when language, which largely resides in the left cerebral hemisphere, dominated and suppressed the images created in the right one.  The idea that brain function before and after literacy could be so different as to bring consciousness of the pre-literate into question was radical enough and hard enough to prove that Jaynes was largely dismissed. Nonetheless, there are tantalizing facts that make his hypothesis interesting.

It has been recognized for more than a hundred years that singing and rhythmic speech are sometimes preserved when the speech areas of the left hemisphere are abruptly destroyed.  In those cases, a different kind of speech originates in the non-dominant side.  One can temporarily put the dominant side of the brain to sleep with barbiturate injected into the left carotid artery.  Why one would do that is a topic for another day, but the result is the same as that seen in a stroke.  Speech is lost but singing and poetry are often preserved.  Even more interesting was an experiment in which positron emission tomograms were done on psychotic patients while they were actively hallucinating.  In those patients neurons in the non-dominant right temporal lobe had increased activity while the visions were happening.  The hallucinations seem to have been generated in the right brain. Ironically, the word paranoia come from the Greek para + nous—literally a mind beside one’s own.  A mind beside one’s own may live in the non-dominant hemisphere, although it rarely surfaces in a brain in which literacy has made the dominant side overwhelmingly important. Perhaps the surfacing was not always rare.

There is a great deal more written about the effect of literacy on brain function, most of which is less controversial than Jaynes.  Whether he was right or not, something about how brains work seems to have changed when civilizations became literate.  In episodes to come, we will come back to how writing evolved and at other ideas about the effect of literacy on the brain.

Reference: Jaynes, Julian, The Origin of Consciousness in the Breakdown of the Bicameral Mind. Boston: Houghton Mifflin Company, 1976.

Communication is central to what we are as humans, and there is a good deal of information that our brains have episodically evolved as we have invented new ways to pass information among ourselves.

                        In the history of hominids there have been four fundamental inventions that changed how we communicate: speech, writing, printing, and coding.  Over the coming months we will look at each and at the science and stories around them.  I should tell you at the outset that this newsletter comes from the research for a book on the topic, and, hopefully, much of what follows will reappear when that project is complete. First, let me summarize the four episodes.

                        The story of speech is perhaps the most difficult to tell since there is no written record of verbal communication’s origin.  Estimates for the beginning of speech range from as little as 60,000 years ago to as much as 1.5 million.  There is also a lively debate about whether speech was a gradually acquired skill or came from a sudden genetic alteration.  There are clues to the answer in the brains of our closest hominin relatives such as chimpanzees and in the skulls and DNA of lost hominids such as Neanderthals and Denisovans.  We will look at all of that.

                        Writing, which began as pictograms around 2500 BCE and progressed to alphabets by 750 BCE, profoundly changed how society and human brains functioned.  We will look at the earliest alphabetical documents—ancient Greek and the Old Testament for example—that were transcriptions of oral tradition that predated writing and carry fascinating clues about how pre-literate brains worked.  In addition, the few surviving pre-literate languages function with only a few thousand words compared to modern English with well over a million and thousands of new ones added every year.  Much can be learned from those cultures, and we will look at them as well.   

                        Printing with movable type dates to 1454 and began the first real information explosion.  More books were printed between 1454 and 1500 than had been copied by hand in the previous 1,000 years.  The Gutenberg press alone printed over 300,000 copies of Martin Luther’s ninety-five theses.  By 1600, 250 European presses had turned out over 12 million books.  The explosion led to libraries, many of which morphed into universities that trained and supported the likes of Newton, Smith, Locke, Montesquieu, Jefferson, and Madison.  The auditory brain became secondary to the visual one, and educated men lost their ability to memorize prodigious amounts of information.  On the positive side, printing made the Renaissance, the Enlightenment, and the Industrial Revolution possible.  New medical science brought antisepsis, anesthesia, and antibiotics and human life span nearly tripled.  Printing was also central to the catastrophic religious wars of the seventeenth century, world wars, genocides, nuclear weapons, and global warming. 

                        The current information explosion began in the 1830s with the first binary code that could be transmitted electronically.  That was followed in less than a century by wireless analog transmission of sound and pictures.  Binary codes reappeared in the 1940s with computers and transistors.  Arpanet linked computers in 1969, and Tim Berners-Lee invented the world wide web in 1989. Now 8.6 trillion texts and 2.5 quintillion bytes of information are exchanged every day.  It is estimated that 90 percent of all the information ever generated by humans was created in the last two years.

                        The human brain has over 100 billion neurons, more than half of which can change either their structure or their connections to meet new demands.  It is hard to imagine that the brain of the current internet adept fifteen-year-old is not profoundly different from that of his sixty-year-old grandfather who grew up with a few radio stations and three television channels.

                        We are just on the cusp of a monumental change in how society and the human brain work, and, if history is any indication, there will be some very good results and some that may not be so good.  We will look in some detail at what happened with the other changes in hopes of better understanding what we are embarking on in the current one. 

                        Join me and have fun.  And I will look forward to your comments, corrections, and additions to what is to come.

Leave a comment

Share The Changing Brain

                        In the history of hominids there have been four fundamental inventions that changed how we communicate: speech, writing, printing, and coding.  Over the coming months we will look at each of them and at their science and stories.  I should tell you at the outset that this newsletter comes from the research for a book on the topic, and, hopefully, much of what follows will reappear when that project is complete. First, let me summarize the four inventions.

                        The story of speech is perhaps the most difficult to tell since there is no written record of verbal communication’s origin.  Estimates for the beginning of speech range from as little as 60,000 years ago to as much as 1.5 million.  There is also a lively debate about whether speech was a gradually acquired skill or came from a sudden genetic alteration.  There are clues to the answer in the brains of our closest hominin relatives such as chimpanzees and in the skulls and DNA of lost hominids such as Neanderthals and Denisovans, and we will look at all of that.

                        Writing, which began as pictograms around 2500 BCE and progressed to alphabets by 750 BCE, profoundly changed how society and human brains functioned.  We will look at the earliest written documents—Homeric epics and the Old Testament for example—that were transcriptions of oral tradition that predated writing and carry fascinating clues about how pre-literate brains worked.  In addition, the few surviving pre-literate languages function with only a few thousand words compared to modern English with well over a million and thousands of new words added every year.  Much can be learned from those cultures, and we will look at them as well.   

                        Printing with movable type dates to 1454 and began the first real information explosion.  More books were printed between 1454 and 1500 than had been copied by hand in the previous 1,000 years.  The Gutenberg press alone printed over 300,000 copies of Martin Luther’s ninety-five theses.  By 1600, 250 European presses had turned out over 12 million books.  The explosion led to libraries, some of which morphed into universities that educated and supported the likes of Newton, Smith, Locke, Montesquieu, Jefferson, and Madison.  The auditory brain became secondary to the visual one, and educated men lost their ability to memorize prodigious amounts of information.  On the positive side, printing made the Renaissance, the Enlightenment, and the Industrial Revolution possible.  New medical science brought antisepsis, anesthesia, and antibiotics, and human life span nearly tripled.  Printing was also central to the catastrophic religious wars of the seventeenth century, two (or maybe three) world wars, multiple genocides, nuclear weapons, and global warming. 

                        The current information explosion began in the 1830s with the first binary code that could be transmitted electronically.  That was followed in less than a century by analog transmission without wires of sound and pictures.  Binary codes came back to the fore in the 1940s with computers and transistors.  Arpanet linked computers in 1969, and Tim Berners-Lee invented the world wide web in 1989. Now 8.6 trillion texts and 2.5 quintillion bytes of information are exchanged every day.  It is estimated that 90 percent of all the information ever generated by humans was created in the last two years.

                        The human brain has over 100 billion neurons, more than half of which can change either their structure or their connections to meet new demands.  It is hard to imagine that the brain of the current internet savvy fifteen-year-old is not profoundly different from that of his sixty-year-old grandfather who grew up with a few radio stations and three television channels.

                        We are just on the cusp of a monumental change in how society and the human brain work, and, if history is any indication, there will be some very good results along with some that are not so good.  We will look in some detail at what happened with the other changes in hopes of better understanding what we are embarking on in the current one. 

                        Join me and have fun. I look forward to your comments, corrections, and additions.